ABSTRACT Wildlife biologists increasingly use camera traps for monitoring animal populations. However, manually sifting through the collected images is expensive and time-consuming. Current

deep learning studies for camera trap images do not adequately tackle real-world challenges such as imbalances between animal and empty images, distinguishing similar species, and the impact

of backgrounds on species identification, limiting the models’ applicability in new locations. Here, we present a novel two-stage deep learning framework. First, we train a global

deep-learning model using all animal species in the dataset. Then, an agglomerative clustering algorithm groups animals based on their appearance. Subsequently, we train a specialized

this strategy using 1.3 million images from 91 camera traps encompassing 24 mammal species and used 120,000 images for testing, achieving an F1-Score of 96.2% using expert models for final

classification. This method surpasses existing deep learning models, demonstrating improved precision and effectiveness in automated wildlife detection. SIMILAR CONTENT BEING VIEWED BY

OTHERS ITERATIVE HUMAN AND AUTOMATED IDENTIFICATION OF WILDLIFE IMAGES Article 18 October 2021 ARTIFICIAL INTELLIGENCE FOR AUTOMATED DETECTION OF LARGE MAMMALS CREATES PATH TO UPSCALE DRONE

SURVEYS Article Open access 18 January 2023 DEEP LEARNING ENABLES SATELLITE-BASED MONITORING OF LARGE POPULATIONS OF TERRESTRIAL MAMMALS ACROSS HETEROGENEOUS LANDSCAPE Article Open access 27

May 2023 INTRODUCTION Camera traps are increasingly used as an effective tool to monitor animal populations in ecological research and conservation1. These automatically triggered sensors

collect images that can be used to conduct fauna inventories, detect elusive species, assess activity patterns, evaluate habitat preferences, estimate occupancy, relative abundance, and

density, and serve for management and dissemination purposes2. Camera traps have become popular in field biology because they allow studying animals remotely and are minimally invasive

compared to traditional capture or marking techniques3. Furthermore, camera traps are often described as a low-cost technology because suitable units that gather thousands of images can be

purchased for around USD200-800 and only require a few periodic field visits to replace memory cards and batteries, which is generally affordable for wildlife research projects. But the

processing of the collected images, which includes visual identification of the targets in the images, labeling, and metadata extraction-and that ultimately leads to data ready for analysis-

constitutes a bottleneck that substantially adds up to the cost of the method and results in a considerable delay in data availability4. There are programs to facilitate this task, e.g.,

for labeling the pictures (e.g., DigiKam (https://www.digikam.org/), Agouti (https://agouti.eu/)) or extracting images’ metadata into datasets (e.g., camtrapR5) (see6 for a review), but

despite these advances, visually identifying the species in the images remains a time-consuming, resource-demanding and tedious task. In this regard, developing Artificial Intelligence (AI)

models for automated species identification represents a dynamic area of ongoing research. Various recent initiatives aim to train AI models using extensive and diverse image datasets7. Some

of these initiatives also offer easy-to-use software platforms and interfaces to streamline the use of AI8,9,10,11. However, identifying animals in camera trap images involves difficulties

for humans and automated systems that stem from factors such as animals too close to the cameras, partially obscured within the images, or subjected to variations in lighting, shadows, and

weather conditions that further complicate the extraction of pertinent information and generate false positives or false negatives in animal detection12, as shown in Fig. 1. More

specifically the most significant challenges for automating animal detection in camera trap projects include: * The imbalance between animal and background: The sparse representation of

animal samples results in an imbalance between the presence of animals and the background. Empty images often result from cameras being triggered by wind moving surrounding vegetation or due

to animals passing by too fast. Consequently, a critical task is to detect whether an image contains an animal (animal vs. empty). * Minimizing background influence on species

identification: The risk of background characteristics affects species identification, for instance, if all wolves consistently appear against a snowy background. In this context, it is

essential to reduce the influence of background features on the precise classification of animal species, thereby preventing misclassifications based on shared background characteristics. *

The imbalance between animal species: Camera trap projects often suffer from significant class imbalance13, wherein certain species are significantly more prevalent than others. The

performance of AI approaches may suffer when models are developed using unbalanced training datasets14. * Differentiating between highly similar animal species: Distinguishing between

closely related animal species (E.g. red deer and fallow deer in Fig. 1) can be challenging, mainly when only a tiny portion of the body, such as a leg, is visible. This challenge persists

even for expert biologists conducting visual identification. This work introduces a two-stage deep learning-based workflow approach (Fig. 2) aimed at automating mammal identification while

addressing the primary challenges encountered in real-world environments. This novel strategy addresses the complexity of animal detection by employing a clustering approach based on groups

of animals determined by species appearance similarity, generating expert models for each group and efficiently decomposing the problem into several simpler sub-problems. Rather than

training a single model encompassing all animal species as different classes, independent models were trained for each animal group, enabling specialization in detecting animals with similar

morphology. In this sense, the concluding phase integrates the global model, trained with all classes across four groups, which redirects its prediction to one of the four expert models to

ascertain the outcome of animal detection. Expert models can achieve improved precision and generalization in detecting animals within their group, benefiting from fewer classes to learn

from and enabling focused analysis of specific features and patterns associated with that particular group. This two-stage approach empowers us to capitalize on the unique strengths of

individual expert models and enhance the final classification outcome based on their respective confidences. This study analyzed a comprehensive five-year image dataset from Sierra de las

Nieves and Doñana National Parks (Andalusia, Spain), comprising 24 mammal species. Our approach, distinct from the conventional deep learning classification methods, leverages an innovative

animal clustering strategy combined with advanced computer vision techniques, specifically MegaDetectorV515 based on YOLOv516. This novel strategy outperforms traditional deep learning

approaches in accuracy, particularly when locating the animal in the image is complex and distinguishing between species with subtle differences. Our results highlight the potential of

combining clustering techniques with deep learning for enhanced object detection in complex natural environments. BACKGROUND AND RELATED WORK Artificial Intelligence (AI) tools have recently

become increasingly important for detecting and identifying objects in images17. Object detection technologies rapidly expand within this approach to detect and localize animals within

images or video frames18. Deep learning-based object detection algorithms are powerfully accurate in finding and locating items in photos, even with complicated backgrounds or when objects

are partially obscured19 and allow automatic animal detection and identification, minimizing manual annotation requirements20. Moreover, a deep learning framework’s detection component

operates more efficiently when anchor boxes are used to predict the location and size of objects in an image21,22. Object detection models focus on learning from the image section

corresponding to the bounding box rather than the entire background. Furthermore, object detection models enhance the interpretability of the results, as they indicate the specific area of

the image where the animal is detected. This information ensures that the model’s classification is based on the animal’s presence rather than the surrounding background. The current object

detectors could be classified into one stage and two stages. One-stage detectors regard object detection as a regression or classification problem and use a unified framework to obtain the

final categories and locations directly23, such as RetinaNet24, Single Shot Detector (SDD)25, AttentionNet426 or You Only Look Once (YOLO). On the contrary, Two-stage detectors generate

regions and classify each area to get different object categories, such as Regions with CNN features (R-CNN)27, Faster Region-based Convolutional Neural Network (Faster R-CNN)28 or

Region-based Fully Convolutional Network (R-FCN)29. One-stage detectors are typically faster and are commonly used for real-time applications. Therefore, One-stage detectors may be more

suitable for animal detection in camera trap photos because of the large number of images to process. Automating animal identification in camera-trap images has been extensively explored

e.g. EventFinder30 software for screening remotely captured images or ClassifyMe31 software for the identification of wildlife in camera trap images. Early works employed Traditional Machine

Learning methods with hand-designed features for animal detection32,33,34. In35, sparse coding spatial pyramid matching (ScSPM) was used to extract local features from camera-trap images,

and a linear Support vector machine (SVM) algorithm was employed for classification. This approach achieved 82% average classification accuracy on a dataset of over 7000 images encompassing

18 species. However, this approach is limited by its dependence on predefined characteristics, which may not adequately capture the extensive variability in wildlife images. Moreover, its

labor-intensive nature impedes scalability and adaptability across diverse ecological settings. In contrast to previous approaches, our work aims to leverage object detection through deep

learning to extract essential features for animal detection automatically. Several recent studies have employed deep learning techniques for camera-trap image classification. For instance,

in36, the authors introduced a two-step process based on Deep Convolutional Neural Networks (DCNN) to classify camera-trap images into three categories: human, animal, and background

patches. Despite using a dataset of 30,000 images for validation, this approach was computationally slow and achieved a recall rate of 73.2%. A different method was proposed in37, utilizing

a DCNN as a feature extractor to train traditional machine learning algorithms like K-nearest neighbors (KNN) and SVM for wildlife animal detection, achieving an accuracy of 91.4% on a

standard camera-trap dataset. Although the dataset contained 20 animal species, each with around 100 image sequences, the classification was limited to animal or background classes. In38,

the authors introduced an automatic computer vision-based species recognition method for camera-trap images. They compiled and annotated a standard camera-trap dataset comprising 20 common

species found in North America, consisting of approximately 20,000 images. Despite their efforts, the achieved accuracy was only 38%, indicating considerable scope for improvement. Many

studies have made use of the Snapshot Serengeti camera-trap dataset (SS hereinafter), which includes images of 48 animal species. In14, the authors assessed the capabilities of

state-of-the-art deep neural networks (DNNs) on the SS dataset. They obtained an accuracy of approximately 57% (estimated from their plot, since the exact accuracy was not explicitly

reported in the paper). Moreover, in39 authors tested DNNs to automatically extract information from images in the standard SS dataset. Remarkably, their method achieved an impressive

overall accuracy of 93.8%. Despite these excellent general accuracy results, the study exhibits limitations in terms of performance when classifying rare classes. Similarly, in40 the

smallest classes had the worst performance (0.18–0.32 F1-Score), while the classifier was robust (0.87–0.95) for highly represented classes. In this regard, one of the most significant

issues is the data imbalance among animal species. For instance, in39, the class imbalance problem is approached by conducting data augmentation through oversampling and employing weighted

loss techniques. The weighted loss method demonstrated the highest top-1 accuracy, with improved classification performance for some rare classes at the expense of reduced accuracy for more

frequent classes. Moreover, the accuracy of the less represented classes increased significantly in some instances while showing no improvement for others. Another significant challenge

encountered in animal image detection using camera traps lies in the discrimination between highly similar species (i.e., red deer and fallow deer) or among groups of species that appear in

similar environmental settings and temporal characteristics. Distinguishing subtle differences between visually similar species is a challenge for both human annotators and machine learning

algorithms, making accurate species classification difficult41. Even experienced biologists may struggle to differentiate between certain species, leading to ambiguous or subjective

labeling42. In animal species classification, some previous work has focused on image classification tasks for which the animals’ species often differ only slightly in small details. For

instance,43,44 work to distinguish different bird species, others such as45 focus on determining moth species with slight variance between classes. These works focus on techniques such as

Fine-grained recognition46, which are part-based solutions that collect additional local information regarding attention or parts. However, these models are sensitive to variations in image

quality, such as lighting conditions, angles, or occlusions, and do not generalize well to low-quality or complex images. In the same way, this challenge is underscored by recent findings47

that shed light on the intricacies associated with discerning among taxonomically related animals using image redundancy contained in sequences of images. This study highlights that the

precision discrepancies between mouflon and goats are largely due to the inherent imbalance in the validation set, where the overrepresentation of chamois and sheep images significantly

skews the classification results. Furthermore, challenges in identifying distinct features, such as ibex horns, contribute to the observed performance issues. Notably, the classification

difficulties extend to nocturnal scenarios, where distinguishing between wolves, foxes, and dogs proves inherently challenging. Thus, this study elucidates the formidable obstacles

associated with accurately classifying closely related species in the context of wildlife monitoring through camera traps. MATERIALS AND METHODS Our work introduces a two-stage deep

learning-based workflow to address the challenges encountered in the literature regarding animal detection from camera trap images. This approach involves an initial Grouping-based strategy

for animal identification, which draws inspiration from field biology, where a generalist initially classifies most images and experts discern highly similar species, enhancing

classification confidence48. The two-stage pipeline outperforms a single 24-class classifier by adopting a divide-and-conquer approach (Fig. 2). In the first stage, the global model

classifies and redirects images to one of four expert models, each with a confidence score. The second stage involves the selected expert model performing its classification. This approach,

focusing on fewer classes, improves precision and generalization by analyzing the distinct features of each group. By grouping similar species, we increased classification accuracy from 92

to 96.2% on a test dataset of over 120,000 images. This strategy enhances overall performance, addressing common issues such as misclassifying empty images and confusing similar species.

DATASET DESCRIPTION The dataset used in this study was collected during fieldwork campaigns in Doñana National Park, Spain (\(36^\circ 59'\) N \(6^\circ 26'\)W) where we deployed

58 camera traps along 2018–2022, and Sierra de las Nieves National Park (\(36^\circ 44'\) N \(4^\circ 59'\)W), Spain where we deployed 33 cameras in 2022 (Fieldwork was conducted

under permits: Doñana National Park 2018/18 and 2021/17; Sierra de las Nieves N. expt.:PNSN/AU/104-2021). Camera traps were attached to trees or wood sticks at a height of 0.5 m above ground

in-game trails and passages where available or in open areas. Whenever a camera trap is triggered, typically in response to nearby animal movement, it captures a series of photographs. In

Doñana, we used the camera model Browning Strike Force HD Pro, and in Sierra de las Nieves, we used a combination of Browning BTC-8E-HP5 Spec Ops Elite HP5 and Browning BTC-5HDPX Strike

Force Pro (https://browningtrailcameras.com/). Additionally, we raised a call for images of species that were underrepresented in our data pull through the Spanish Society for the

Conservation and Study of Mammals (SECEM (https://www.secem.es/)) and received complementary images from citizens and colleagues of several locations in Spain, those obtained with various

camera models. The dataset employed in this study comprises 1,331,309 images encompassing 24 distinct species: wild boar _Sus scrofa_; feral cattle _Bos taurus_; feral horse _Equus

caballus_; domestic sheep _Ovis orientalis aries_; domestic goat _Capra aegagrus hircus_; roe deer _Capreolus capreolus_; Iberian ibex _Capra pyrenaica_; European mouflon _Ovis aries

musimon_; red deer _Cervus elaphus_; fallow deer _Dama dama_; red fox _Vulpes vulpes_, Egyptian mongoose _Herpestes ichneumon_, Eurasian badger _Meles meles_, feral dog _Canis lupus

familiaris_; feral cat _Felis silvestris catus_; small-common genet _Genetta genetta_, Eurasian otter _Lutra lutra_; Iberian lynx _Lynx pardinus_; European wildcat _Felis silvestris_; stone

marten _Martes foina_; rabbit _Oryctolagus cuniculus_; hare _Lepus granatensis_; mouse _Mus_ or _Apodemus sp._; rat _Rattus rattus_. ACCELERATING THE ANNOTATION PROCESS OF CAMERA TRAP IMAGES

Deep learning algorithms based on object detection techniques have shown promising results in camera trap animal identification. However, the animal must be located and labeled inside the

image to train these algorithms. Annotation includes tagging the image and generating bounding boxes, which involves enormous manual labor. In our research, we leveraged the capabilities of

MegaDetectorV515, based on the YOLOv5 architecture, as a semi-automatic labeling tool for detecting animals in camera trap images. Developed by Microsoft, this versatile object detection

model is specifically designed to analyze extensive camera trap datasets. Trained on millions of images from diverse global environments, MegaDetector can identify three primary object

classes: humans, animals, and vehicles. It also efficiently recognizes empty images that do not contain these specific classes. The use of MegaDetector substantially accelerates the

annotation process, enabling biological experts to bypass the tedious manual task of locating animals within each image. Instead, experts can direct their efforts towards more nuanced

aspects of data analysis. This semi-automatic approach to generate bounding boxes around detected animals streamlines the workflow, enhancing overall efficiency. In this sense, several

research studies, such as11,49,50, have assessed the efficiency of MegaDetector in wildlife detection, e.g.,11 found that its use increased the processing speed by over 500%. The time

required for the manual processing component was reduced by 8.4 times. In total, 596, 748 photos have been annotated with this semi-automatic labeling approach. The remaining images were

labeled as _empty_, indicating the absence of animals. Moreover, the preparation of training data carefully accounted for the diverse range of environmental factors and capturing conditions

of a real-world use case. Therefore, we aimed to make the data as diverse as possible, including photos from different times of day, seasons, weather conditions, lighting conditions, angles,

and camera setups. This approach helps to enhance the model’s robustness and adaptability, ultimately leading to more accurate and reliable animal detection and classification in real-world

camera trap settings. GROUPING-BASED STRATEGY FOR ANIMAL IDENTIFICATION We have developed an innovative clustering approach based on the animal’s appearance, which generates expert models

for each group. Rather than training a single model encompassing all animal species as different classes, independent models were trained for each animal group, enabling specialization in

detecting animals with similar morphology. We employed a systematic approach that integrates machine learning techniques to automate the grouping of animal species based on their detection

similarities (Fig. 3). The first step involves the construction of a confusion matrix, denoted as _C_, which captures the performance of our detection models by revealing the count of

correct and misclassified predictions for each species. Let _N_ be the total number of species under consideration, and \(C_{ij}\) represent the count of instances where species _i_ was

predicted as species _j_. To normalize this matrix and derive a similarity matrix, _S_, each element \(S_{ij}\) is computed as: $$\begin{aligned} S_{ij} = \frac{C_{ij}}{\sum _{k=1}^{N}

C_{ik}}. \end{aligned}$$ (1) The resulting _S_ matrix represents the normalized similarity between species, where values close to 1 indicate high similarity and values close to 0 indicate

dissimilarity. With the similarity matrix in hand, we applied hierarchical agglomerative clustering to group species accordingly. The linkage matrix resulting from this process provides

insights into the hierarchical structure of species relationships. By visualizing the dendrogram derived from the linkage matrix, we obtain a tree-like representation, where closely related

species happened to appear in proximity. To facilitate interpretability and application, we determined the number of clusters (_K_) based on the dendrogram structure four in our case.

Subsequently, the agglomerative clustering algorithm was employed to assign each species to one of the _K_ clusters. This process results in an assignment vector \(\textbf{A}\), where

\(A_i\) represents the cluster to which species _i_ is assigned. It is remarkable that this automated process produced clusters that perfectly aligned with the species’ biological

characteristics, such as taxonomy, size, behavior, and ecological roles and where therefore validated by the biologists: * 1. Large mammals: Order Perissodactyla (Fam. Equidae) and Order

Artiodactila (Fam. Bovidae and Suidae) * 2. Medium mammals: Order Carnivora (Fam. Canidae and Felidae) * 3. Small mammals: Order Carnivora (Fam. Mustelidae, Viverridae, and Herpestidae) * 4.

Micromammals: Order Lagomorpha (Fam. Leporidae) and Order Rodentia (Fam. Muridae) DEEP LEARNING ARCHITECTURES In this work, various state-of-the-art Deep Neural Networks were tested and

compared to identify the highest-performing networks. Specifically, deep learning models focused on image classification, such as GoogLeNet51, AlexNet52, ResNet5053, ResNet10153,

ResNet15253, VGG1654, VGG1954, and Xception55, were evaluated. Additionally, deep learning-based object detection models, including YOLOv5, YOLOv5 (MegaDetector), YOLOv8, and Faster R-CNN,

were also tested. Each deep learning model was trained with all animal classes in the training set and evaluated using the F-score metric (eq 3), which balances precision and recall (Eq. 2).

$$\begin{aligned} precision = \frac{TP}{TP + FP} \nonumber \\ \quad recall = \frac{TP}{TP + FN} \end{aligned}$$ (2) $$\begin{aligned} F1-score = \frac{2 * precision * recall}{precision +

recall} \end{aligned}$$ (3) Precision measures the accuracy of positive predictions, while recall measures the ability to identify all relevant instances. The F-score, as the harmonic mean

of precision and recall, comprehensively evaluates the model’s performance. Relying solely on accuracy can be misleading, particularly with imbalanced datasets, as it may overemphasize

dominant classes and neglect rare or endangered species. The F-score offers a more nuanced assessment, ensuring that performance across all classes is fairly evaluated and that more frequent

ones do not overshadow the detection of less common species. This ensures a robust evaluation of the model’s effectiveness in real-world applications. As shown in Table 1, the best results

were obtained using YOLOv5 and MegaDetector as pre-trained models, achieving an F-score of 92%. Transfer Learning56, a widely adopted technique, enables the reuse of knowledge gained from

one task to serve as a starting point for training models on related tasks. Accordingly, we used MegaDetector as a pre-trained model in our two-stage pipeline, leveraging the extensive

knowledge it acquired from diverse training on a wide range of animal species and ecosystems. More details about model architecture and hyperparameters can be found in Appendix A:

Experimental settings. ENHANCING ANIMAL DETECTION WITH A TWO-STAGE PIPELINE USING EXPERT MODELS Finally, a global model was trained with all animal classes and four expert models for each

group of animals. Table 2 shows how each expert model achieves improved metrics on the validation sets by reducing the number of species per model and considering similar species within each

model. Overall, the expert models exhibit superior animal detection and classification performance in the images. The global model tends to fail in classifying large mammals. This is often

because these animals appear close to the camera trap, resulting resulting in only a small part of the body being visible. Additionally, the global model struggles with detecting

micromammals, as these animals are typically difficult to spotin the images due to their small size. RESULTS A series of experiments were conducted to demonstrate the feasibility of our

proposal on an out-of-sample test dataset containing over 120,000 images. These images were taken in contexts that were never encountered during the training stage. In the first stage of the

methodology, the global model classifies images into four groups, obtaining an impressive F-score of 97% (Fig. 4). This high performance is due to the ease with which the global model can

distinguish between these four major groups of animals, whose appearance characteristics are significantly different. In the second stage, the decision from the global model is redirected to

the expert models, achieving a final F-score of 96.2%. Next, we will showcase how our approach achieves enhanced accuracy in addressing the challenges of animal detection in real-world

scenarios. We will compare the metrics obtained from a single-stage pipeline, which uses a global model trained on all classes, with the metrics from our two-stage pipeline (Table 3). Our

two-stage approach demonstrates significant improvements in accuracy due to the use of models customized for groups of animals with similar morphological characteristics. For a more detailed

view of the results, see Appendix B: Two-stage pipeline evaluation. CHALLENGE 1: DEALING WITH BACKGROUND Most models implemented using a one-stage pipeline tend to misclassify many images

as empty, particularly in challenging conditions such as low light, nighttime, or when the animal is partially concealed. In this scenario (Fig. 2 scenario B), when the global model

classifies the image as “background”, the response is redirected to all expert models, which then make the final decision based on their confidence rates. This scenario is the most complex

in the decision-making process, as it is where the global model tends to have a higher error rate (see global confusion matrix). There are two crucial situations in these cases: * True

Positive (Background): The global model predicts “background,” and the image is indeed empty. In this case, the global model redirects the response to all four expert models, which also

predict “background” 100% of the time. * False Negative (Animal): The global model predicts“background,” but the image actually contains an animal. Here, the response is redirected to the

four expert models. Typically, all expert models detect the presence of an animal, even if it does not belong to their respective groups. The final decision is made by the expert model with

the highest confidence rate. The model containing the animal usually has a very high confidence rate, while the other three models, which do not contain the animal in their groups, typically

have confidence rates below 60%. This approach has demonstrated the capability to identify the animal in the image in over 99% of cases, including those where the animal is hidden far in

the background, such as large mammals like _feral cattle (bos)_, _red deer (cer)_, and _feral horse (equ)_, or micromammals like _mouse (mus)_ that are very difficult to identify due to

their small size. See Appendix B: Two-stage pipeline evaluation for more details. CHALLENGE 2: DEALING WITH UNBALANCED DATASETS The dataset is heavily unbalanced, with some species being

much more frequent than others. This imbalance poses challenges for machine learning techniques, which tend to be biased towards classes with more examples. Consequently, the model may

primarily predict the more frequent types, such as _wild boar_ or _red deer_, achieving high accuracy without effectively learning the less common classes. This is particularly concerning

when less frequent classes like the _Iberian lynx_ hold greater scientific interest and conservation importance. Addressing class imbalance is, therefore, crucial to ensure comprehensive and

accurate ecological studies. The Grouping-Based Strategy mitigates class imbalance by reducing the number of animal classes each expert model needs to handle. Notably, animals with similar

characteristics often exhibit analogous patterns, resulting in comparable appearances in camera trap images. For instance, larger mammals tend to have a higher frequency of occurrences in

the dataset than smaller mammals. Thus, our strategy, grounded in animal grouping, presents a promising solution to the inherent challenge of imbalanced data. As an example, our approach has

improved the F-score of minority classes (see Table 3), such as the _Iberian lynx (lyn)_ from 76% to 93%, _roe deer (caca)_ from 57 to 95% , _fallow deer (dam)_ from 38 to 82% or _European

wildcat (fel)_ from 70 to 93%. It is important to note that class balancing strategies were implemented during training for each animal group model since there were some less-represented

species within each group. However, these data-balancing strategies become more straightforward and effective as the number of classes decreases. To address this imbalance, we modified the

YOLO configuration file (typically in YAML format) to assign higher weights to underrepresented species in the dataset. Additionally, we applied data augmentation techniques targeted at the

most imbalanced species. This approach helped improve overall performance and mitigated the risk of the model becoming biased towards species with larger sample sizes. CHALLENGE 3: DEALING

WITH SIMILAR ANIMAL SPECIES Most misclassifications in one-stage models trained with all classes occur among very similar species, such as _roe deer_ and _fallow deer_, _rabbit_ and _hare_,

or _Iberian lynx_ and other felines. Additionally, misclassifications occur with some larger mammals (e.g., _feral cattle_, _feral horse_), particularly in night images where species

identification is inherently challenging or when only a small portion of the animal, such as a leg or hoof, is visible. Thanks to the grouping based on appearance and size, we can develop

more specialized models that mitigate the bias from highly similar species. By clustering animals with similar physical characteristics, our approach allows each expert model to focus on a

narrower set of species, enhancing its ability to differentiate between them accurately. This specialization reduces the confusion often observed in models trained on a wide variety of

species. For instance, when species such as _red deer and_ _fallow deer_ are grouped based on their size and appearance, the expert model can fine-tune its parameters to recognize subtle

differences that a general model might overlook, increasing the F-score from 86 to 92% for _red deer_ and from 38 to 82% for _fallow deer_, as shown in Table 3. Similarly, by clustering

smaller animals like rabbits and hares, the F-score improved from 77 to 94% and 95 to 97%, respectively. For larger mammals such as _feral cattle_ and _feral horse_, the F-score increased

from 88 to 94% for both species. Each model becomes adept at identifying species’ unique features within its group. This targeted approach improves the overall accuracy of species

identification and addresses the common issue of misclassification in one-stage models. It ensures that the specialized models maintain high performance even in challenging conditions, such

as night images or when only a part of the animal is visible. By reducing the workload on a single, all-encompassing model, our strategy promotes more reliable and precise animal detection

across diverse ecological studies. For a more detailed view of the results, see Appendix B: Two-stage pipeline evaluation. DISCUSSION Our study introduces the first AI workflow for object

detection in camera trap images that employs a serial approach, combining a general model followed by expert models tailored for specific animal groups. While current deep learning models

have significantly improved detection capabilities, developing algorithms that perform well across various classes remains challenging. Most works in this area extend multi-class models to

handle multiple types, but these models often need more flexibility when accommodating new object descriptors. Furthermore, they typically do not address multiple criteria simultaneously.

Some approaches rely on iteratively executing single-class models for each target class, which results in linear scaling of training and run times. Our serial approach has immediate

applicability to solve real-world challenges in camera-trap scenarios, and its main advantages for producing higher-quality results are: * Reducing the data unbalance problem. Clustering

animals in groups reduces the significance of sample differences between classes. * Reduce misclassification among similar species. By training expert models for specific animal groups, the

models can focus on more subtle details that allow for distinguishing similar species. This is a highly relevant result because classifying animal species correctly is crucial.

Misclassification can result in erroneous fauna inventories and biases in estimating animal species’ geographic range or habitat use, negatively affecting scientific studies’ reliability and

hindering conservation and management efforts48. Misidentification is common among phenotypically similar species, even for trained human observers42,48. For example,42 found that none out

of ten human observers assessed for identifying animals in camera trap images identified all mammalian wildlife across the sub-datasets and that the majority of misidentification happened

between similar species (two African rhinoceros species, two zebra species as well and two similar species of dwarf antelopes (n = 12, 52.2%). Similarly,48 found that experts were sometimes

even inconsistent with themselves, providing different classifications for similar species (bobcat or lynx) over the same images shown in two trials separated by ten weeks. * Reduce the

number of false negatives. Expert models can learn subtle characteristics of the animals belonging to specific groups.42 found that the species missed mainly by human observers were small

mammals (56.5%) with body masses < 5 kg. For example, it is common that the model fails to detect mouse in the image, as these animals are often well-hidden and typically appear in photos

taken at night. However, independent small mammal models can detect mouse even in complicated situations. Although there are many studies on multi-class animal species detection, creating

algorithms that work well with multiple species remains challenging. Designing efficient multi-class species detection systems is a complex and active research area, as the scale of species

classes varies depending on the tasks. Our two-stage pipeline can be a solution for creating a scalable multi-class model with many classes. Creating a global model to distinguish between

groups of animals is both straightforward and effective. The model’s performance and taxonomic coverage suggest its potential for automatically sorting vast quantities of images across

various taxonomic groups. This capability is particularly valuable in studies examining the impact of anthropization on large mammal communities47. Furthermore, our semi-automatic clustering

approach for developing expert models for each animal group reduces the workload on a single model with a high number of classes, resulting in promising improvements in overall performance.

LIMITATIONS AND FUTURE WORK The methodology presented here is based on algorithms trained using manually labeled images. Therefore, any human error in the classification used as“truth”

would be introduced to the AI workflow. To avoid or minimize it, we emphasize the importance of using expert and multiple observers, as recommended by42,48 and that these observers discard

the images in the training dataset where species cannot be confidently classified. Our approach has room for improvement in future developments. In the current case, the four groups into

which all species were divided were created using our grouping-based strategy, with the number of groups decided by expert biologists. This could be accomplished in the future using other

clustering algorithms based on Artificial Intelligence, allowing for a fully automatic methodology that can be extrapolated to other scenarios with different species compositions. DATA

AVAILIBILITY The datasets generated and/or analyzed during the current study are not publicly available due to its large volume, but are available from the corresponding author on reasonable

request. REFERENCES * Rovero, F., Kays, R. Camera trapping for conservation. In _Conservation Technology_, chap. 10 (eds. Wich, S. & Piel, A. K.) 79–101 (Oxford University Press, 2021).

* Boitani, L. _Camera Trapping for Wildlife Research_ (Pelagic Publishing Ltd., 2016). * Rovero, F., Tobler, M. & Sanderson, J. Camera trapping for inventorying terrestrial vertebrates.

Manual on field recording techniques and protocols for all taxa biodiversity inventories and monitoring. _Belgian Natl. Focal Point Glob. Taxon. Initiat._ 8, 100–128 (2010). Google Scholar

  * Norouzzadeh, M. S. et al. Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. _Proc. Natl. Acad. Sci._ 115, E5716–E5725.

https://doi.org/10.1073/pnas.1719367115 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Niedballa, J., Sollmann, R., Courtiol, A. & Wilting, A. camtrapr: an r package for

efficient camera trap data management. _Methods Ecol. Evol._ 7, 1457–1462 (2016). Article  Google Scholar  * Young, S., Rode-Margono, J. & Amin, R. Software to facilitate and streamline

camera trap data management: A review. _Ecol. Evol._ 8, 9947–9957 (2018). Article  PubMed  PubMed Central  Google Scholar  * Vélez, J. et al. An evaluation of platforms for processing

camera-trap data using artificial intelligence. _Methods Ecol. Evol._ 14, 459–477 (2023). Article  Google Scholar  * Hendry, H., Mann, C. Camelot–intuitive software for camera trap data

management. _BioRxiv_ 203216 (2017). * Chalmers, C., Fergus, P., Wich, S., Montanez, A.C. Conservation ai: Live stream analysis for the detection of endangered species using convolutional

neural networks and drone technology. arXiv preprint arXiv:1910.07360 (2019). * Tabak, M. A. et al. Improving the accessibility and transferability of machine learning algorithms for

identification of animals in camera trap images: Mlwic2. _Ecol. Evol._ 10, 10374–10383 (2020). Article  PubMed  PubMed Central  Google Scholar  * Fennell, M., Beirne, C. & Burton, A. C.

Use of object detection in camera trap image identification: Assessing a method to rapidly and accurately classify human and animal detections for research and application in recreation

ecology. _Glob. Ecol. Conserv._ 35, e02104 (2022). Google Scholar  * Peng, J. et al. Wild animal survey using uas imagery and deep learning: modified faster r-cnn for kiang detection in

tibetan plateau. _ISPRS J. Photogramm. Remote. Sens._ 169, 364–376 (2020). Article  ADS  Google Scholar  * Zhu, H., Tian, Y. & Zhang, J. Class incremental learning for wildlife

biodiversity monitoring in camera trap images. _Eco. Inform._ 71, 101760 (2022). Article  Google Scholar  * Villa, A. G., Salazar, A. & Vargas, F. Towards automatic wild animal

monitoring: Identification of animal species in camera-trap images using very deep convolutional neural networks. _Eco. Inform._ 41, 24–32 (2017). Article  Google Scholar  * Beery, S.,

Morris, D. & Yang, S. Efficient pipeline for camera trap image review. 1907, 06772 (2019). * Jocher, G. _et al._ ultralytics/yolov5: v3.0. Zenodo (2020). * Hughey, L. F., Hein, A. M.,

Strandburg-Peshkin, A. & Jensen, F. H. Challenges and solutions for studying collective animal behaviour in the wild. _Philos. Trans. R. Soc. B Biol. Sci._ 373, 20170005 (2018). Article

  Google Scholar  * Wu, X., Sahoo, D. & Hoi, S. C. Recent advances in deep learning for object detection. _Neurocomputing_ 396, 39–64 (2020). Article  Google Scholar  * Tong, K., Wu, Y.

& Zhou, F. Recent advances in small object detection based on deep learning: A review. _Image Vis. Comput._ 97, 103910 (2020). Article  Google Scholar  * Guo, Y. et al. Varied channels

region proposal and classification network for wildlife image classification under complex environment. _IET Image Proc._ 14, 585–591 (2020). Article  Google Scholar  * Gao, M., Du, Y.,

Yang, Y. & Zhang, J. Adaptive anchor box mechanism to improve the accuracy in the object detection system. _Multimed. Tools Appl._ 78, 27383–27402 (2019). Article  Google Scholar  *

Miao, Z. _et al._ Insights and approaches using deep learning to classify wildlife. _Sci. Rep._ 9(1), 1–9 (2019). * Zhao, Z.-Q., Zheng, P., Xu, S.-T. & Wu, X. Object detection with deep

learning: A review. _IEEE Trans. Neural Netw. Learn. Syst._ 30, 3212–3232 (2019). Article  PubMed  Google Scholar  * Lin, T.-Y., Goyal, P., Girshick, R., He, K., Dollár, P. Focal loss for

dense object detection. In _Proceedings of the IEEE International Conference on Computer Vision_, 2980–2988 (2017). * Liu, W. _et al._ Ssd: Single shot multibox detector. In _Computer

Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14_, 21–37 (Springer, 2016). * Yoo, D., Park, S., Lee, J.-Y., Paek, A.S., So

Kweon, I. Attentionnet: Aggregating weak directions for accurate object detection. In _Proceedings of the IEEE International Conference on Computer Vision_, 2659–2667 (2015). * Girshick, R.,

Donahue, J., Darrell, T., Malik, J. Rich feature hierarchies for accurate object detection and semantic segmentation. In _Proceedings of the IEEE Conference on Computer Vision and Pattern

Recognition_, 580–587 (2014). * Ren, S., He, K., Girshick, R., Sun, J. Faster r-cnn: Towards real-time object detection with region proposal networks. _Advances in Neural Information

Processing Systems_ 28 (2015). * Dai, J., Li, Y., He, K., Sun, J. R-fcn: Object detection via region-based fully convolutional networks. _Adv. Neural Inf. Process. Syst._ 29 (2016). *

Janzen, M., Ritter, A., Walker, P. D. & Visscher, D. R. Eventfinder: a program for screening remotely captured images. _Environ. Monit. Assess._ 191, 1–10 (2019). Article  CAS  Google

Scholar  * Falzon, G. et al. Classifyme: a field-scouting software for the identification of wildlife in camera trap images. _Animals_ 10, 58 (2019). Article  PubMed  PubMed Central  Google

Scholar  * Parikh, M., Patel, M. & Bhatt, D. Animal detection using template matching algorithm. _Int. J. Res. Mod. Eng. Emerg. Technol_ 1, 26–32 (2013). Google Scholar  * Swinnen, K.

R., Reijniers, J., Breno, M. & Leirs, H. A novel method to reduce time investment when processing videos from camera trap studies. _PLoS One_ 9, e98881 (2014). Article  ADS  PubMed 

PubMed Central  Google Scholar  * Antônio, W. H., Da Silva, M., Miani, R. S. & Souza, J. R. A proposal of an animal detection system using machine learning. _Appl. Artif. Intell._ 33,

1093–1106 (2019). Article  Google Scholar  * Yu, X. et al. Automated identification of animal species in camera trap images. _EURASIP J. Image Video Process._ 2013, 1–10 (2013). Article  ADS

  Google Scholar  * Yousif, H., Yuan, J., Kays, R., He, Z. Fast human-animal detection from highly cluttered camera-trap images using joint background modeling and deep learning

classification. In _2017 IEEE International Symposium on Circuits and Systems (ISCAS)_, 1–4 (IEEE, 2017). * Verma, G.K., Gupta, P. Wild animal detection using deep convolutional neural

network. In _Proceedings of 2nd International Conference on Computer Vision & Image Processing: CVIP 2017_, vol. 2, 327–338 (Springer, 2018). * Chen, G., Han, T.X., He, Z., Kays, R.,

Forrester, T. Deep convolutional neural network based species recognition for wild animal monitoring. In _2014 IEEE International Conference on Image Processing (ICIP)_, 858–862 (IEEE,

2014). * Norouzzadeh, M. S. et al. Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. _Proc. Natl. Acad. Sci._ 115, E5716–E5725

(2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Zualkernan, I. et al. An iot system using deep learning to classify camera trap images on the edge. _Computers_ 11, 13 (2022).

Article  Google Scholar  * Zhao, B., Feng, J., Wu, X. & Yan, S. A survey on deep learning-based fine-grained object classification and semantic segmentation. _Int. J. Autom. Comput._

14, 119–135 (2017). Article  Google Scholar  * Zett, T., Stratford, K. J. & Weise, F. J. Inter-observer variance and agreement of wildlife information extracted from camera trap images.

_Biodivers. Conserv._ 31, 3019–3037 (2022). Article  Google Scholar  * Korsch, D., Bodesheim, P., Denzler, J. Classification-specific parts for improving fine-grained visual categorization.

In _Pattern Recognition: 41st DAGM German Conference, DAGM GCPR 2019, Dortmund, Germany, September 10–13, 2019, Proceedings 41_, 62–75 (Springer, 2019). * Cui, Y., Song, Y., Sun, C., Howard,

A., Belongie, S. Large scale fine-grained categorization and domain-specific transfer learning. In _Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition_, 4109–4118

(2018). * Rodner, E. et al. Fine-grained recognition datasets for biodiversity analysis. 1507, 00913 (2015). * Gebru, T., Hoffman, J., Fei-Fei, L. Fine-grained recognition in the wild: A

multi-task domain adaptation approach. In _Proceedings of the IEEE International Conference on Computer Vision_, 1349–1358 (2017). * Rigoudy, N. et al. The deepfaune initiative: a

collaborative effort towards the automatic identification of European fauna in camera trap images. _Eur. J. Wildl. Res._ 69, 113 (2023). Article  Google Scholar  * Gooliaff, T. & Hodges,

K. E. Measuring agreement among experts in classifying camera images of similar species. _Ecol. Evol._ 8, 11009–11021 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Leorna,

S., Brinkman, T. Human vs. machine: Detecting wildlife in camera trap images. _Ecol. Inform._ 72, 101876 (2022). * VÃlez, J. _et al._ Choosing an appropriate platform and workflow for

processing camera trap data using artificial intelligence (2022). 2202.02283. * Szegedy, C. _et al._ Going deeper with convolutions. In _Proceedings of the IEEE Conference on Computer Vision

and Pattern Recognition_, 1–9 (2015). * Krizhevsky, A., Sutskever, I., Hinton, G.E. Imagenet classification with deep convolutional neural networks. _Adv. Neural Inf. Process. Syst._ 25

(2012). * He, K., Zhang, X., Ren, S., Sun, J. Deep residual learning for image recognition. In _Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition_, 770–778

(2016). * Simonyan, K., Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014). * Chollet, F. Xception: Deep learning with

depthwise separable convolutions. In _Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition_, 1251–1258 (2017). * Lumini, A. & Nanni, L. Deep learning and

transfer learning features for plankton classification. _Eco. Inform._ 51, 33–43 (2019). Article  Google Scholar  * Deng, J. _et al._ Imagenet: A large-scale hierarchical image database. In

_2009 IEEE Conference on Computer Vision and Pattern Recognition_, 248–255 (IEEE, 2009). Download references ACKNOWLEDGEMENTS This work has been partially funded by: Project RNM262-G-FEDER

funded by Plan Propio Universidad de Málaga, Junta de Andalucía, Spain, FEDER granted to R.R.; Project AETHER-UMA PID2020-112540RB-C41 funded by MCIN/AEI/ 10.13039/501100011033 granted to

S.H., C.B., I.N. and J.F.A.; LifeWatch ERIC Project _EnBiC2-Lab_, co-funded by the ERDF (Spain’s Pluri-regional Operative Programme 2014-2020) through the Spanish Ministry for Research and

Innovation granted to R.R and J.A.M; Project QUAL21 010UMA funded by Junta de Andalucía, Spain granted to S.H.; Project PY20-01063 funded by the Junta de Andalucía through the Andalusian

Plan for Research, Development, and Innovation (PAIDI 2020) co-financed by the European Union FEDER Andalusia Operational Program 2014-2020 granted to R.R.; Project B1-2022_09 funded by

University of Málaga (B1.- Young researchers projects. II Plan Propio de investigación y transferencia. 2022) granted to M.M.P.; Early Career Researchers Seed Corn Funding and ECR Fellowship

2018-22 funded by Liverpool John Moores University, both granted to M.M.P. M.M.P. Research fellow contract was funded by Ministerio de Universidades, Spain, European Union NextGenerationEU

María Zambrano program 2021-23. F.D.R. was supported by a postdoctoral research contract funded by University of Málaga through the grant program Ayudas para la Incorporación de Doctores del

I Plan Propio de Investigación de la Universidad de Málaga (Call 2020). We thank the Spanish Society for the Conservation and Study of Mammals SECEM for dissemination of the call request

for images and the people who kindly provided them: Luis Javier Palomo; José Guerrero Casado; Javier Fernandez de Simon; Joan Ander Zearra; David Relimpio; Christian Gortázar; Carlos Sanz;

Vincenzo Penteriani; Gabi Berasategui; Jesús Caro; Federico Ojeda; Javier Gómez. We thank Doñana Biological Reserve CSIC and ICTS for logistic support in Doñana NP and Jose María Galán

(Espacio Natural Doñana) for his help on species identification on camera trap images. AUTHOR INFORMATION Author notes * Margarita Mulero-Pázmány and Sandro Hurtado contributed equally to

this work. AUTHORS AND AFFILIATIONS * Department of Animal Biology, University of Málaga, 29071, Málaga, Spain Margarita Mulero-Pázmány, Francisco Díaz-Ruiz & Raimundo Real * KHAOS

Research Group, ITIS Software, University of Málaga, 29071, Málaga, Spain Sandro Hurtado, Cristóbal Barba-González, María Luisa Antequera-Gómez, Ismael Navas-Delgado & José F.

Aldana-Montes Authors * Margarita Mulero-Pázmány View author publications You can also search for this author inPubMed Google Scholar * Sandro Hurtado View author publications You can also

search for this author inPubMed Google Scholar * Cristóbal Barba-González View author publications You can also search for this author inPubMed Google Scholar * María Luisa Antequera-Gómez

View author publications You can also search for this author inPubMed Google Scholar * Francisco Díaz-Ruiz View author publications You can also search for this author inPubMed Google

Scholar * Raimundo Real View author publications You can also search for this author inPubMed Google Scholar * Ismael Navas-Delgado View author publications You can also search for this

author inPubMed Google Scholar * José F. Aldana-Montes View author publications You can also search for this author inPubMed Google Scholar CONTRIBUTIONS M.M.P.: Conceptualization,

methodology, investigation, data curation, writing—original draft, writing—review and editing, project administration, funding acquisition. S.H: Conceptualization, methodology, software,

writing—original draft, writing—review and editing, supervision, visualizations. C.B: Conceptualization, methodology, writing—original draft, writing—review and editing, supervision, project

administration. M.A: Conceptualization, methodology, investigation, data curation, writing—review and editing, project administration. F.D: Writing—Review and Editing. R.R: Writing—review

& editing, funding acquisition. I.N: Writing—review & editing, funding acquisition. J.A: Review and editing, funding acquisition. CORRESPONDING AUTHOR Correspondence to Margarita

Mulero-Pázmány. ADDITIONAL INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. APPENDIX

APPENDIX APPENDIX A: EXPERIMENTAL SETTINGS Transfer learning is a powerful technique that leverages knowledge gained from one task and applies it to a different, related study. In

particular, we used GoogLeNet, AlexNet, ResNet50, ResNet101, ResNet152, VGG16, VGG19, and Xception as pre-trained models on the ImageNet dataset, which contains 1.3 million images from 1,000

classes of man-made and natural images57. These models were employed to extract high-level features from the images. Subsequently, we trained a classifier on top of these features using our

dataset to classify all animal species in Doñana and Sierra de la Nieves study areas. We trained the networks via backpropagation using Stochastic Gradient Descent (SGD) optimization with

momentum and weight decay. Each model underwent training for 100 epochs, starting with an initial learning rate of 0.01. The model was checkpointed after each epoch, and the results of the

most accurate model on the expert-labeled test set were reported. The settings for these experiments are detailed in _Deep Learning Classification_ Table 4. Similarly, we configured the

training parameters for the deep learning object detection models, including YOLOv5, YOLOv5 (MegaDetector), YOLOv8, and Faster R-CNN, as shown in _Deep Learning Detection_ Table 4. Finally,

we used YOLOv5 with the MegaDetector model as a pre-trained model. The MegaDetector model has been trained on a large dataset comprising millions of images captured across diverse locations

worldwide. For our specific task, we used the MegaDetector model as a starting point, initializing the weights of our models for further training on our dataset. By freezing the model’s

first ten layers (layers 0-9), representing the backbone architecture, we can extract high-level features from the millions of camera trap images. These features are then used to classify

our animal species, effectively taking advantage of the rich knowledge learned by the MegaDetector model during its extensive training on a diverse range of ecosystems. Deep Learning

detection models were trained using backpropagation, implementing Stochastic Gradient Descent (SGD) optimization with momentum and weight decay. Each model underwent training for 300 epochs,

starting with an initial learning rate and a weight decay policy, as shown in Table 4. To ensure robustness, we used a batch size of 128 with an image size of 1280, facilitating effective

batch normalization of the input images. Throughout the training process, we took periodic snapshots of the model after each epoch, enabling us to track its progress. Additionally, we

implemented a model checkpoint with patience of 50 epochs, meaning that if the model’s accuracy did not improve within 50 epochs, we reverted to the best-performing model. Finally, we

evaluated and reported the results of the most accurate model based on its performance on the validation set. This approach allowed us to identify and select the best model for further

analysis and evaluation. APPENDIX B: TWO-STAGE PIPELINE EVALUATION Measured on the labeled test set, the global model achieved an F1-score of 0.92%. Despite achieving good results, we have

identified several challenges that must be addressed, as shown in Fig. 5. The model accurately classifies images without animals as background images, which makes sense given the pre-trained

MegaDetector model. However, the model struggles to detect certain animals in specific situations, such as smaller animals like mouse. Furthermore, the model tends to confuse similar animal

species, such as deer and roe deer or hare and rabbit. Our two-stage pipeline leverages the strengths of both a global model and specialized expert models. The confusion matrix of our

two-stage pipeline, shown in Figure 6, demonstrates significant improvements in accuracy and classification performance. * The specialized expert models are trained to focus on specific

groups of animals, allowing for better detection of smaller species such as mouse. The system enhances the identification accuracy of these species by redirecting the classification task to

an expert model that is fine-tuned for small mammals. * Our approach reduces the confusion between similar species by employing expert models specializing in specific morphological

characteristics. For example, red deer and fallow deer, and hare and rabbit are more accurately classified by models trained on features specific to each group. This specialization minimizes

misclassification errors and improves overall precision. * The two-stage pipeline mitigates the confusion caused by shared characteristics among different species. The global model’s

initial classification step ensures images are directed to the appropriate expert model. It can then use its specialized knowledge to distinguish between species that may appear similar when

only partial features are visible, such as the legs of cows and horses. * While the global model performs well in identifying background images, the two-stage approach further refines this

by using the confidence scores from expert models to confirm the absence of animals. This ensures that images are accurately classified as background, reducing false positives. Moreover,

Figs. 7 and 8 illustrate examples where our two-stage pipeline based on expert models significantly improves animal species classification. In these challenging scenarios, the global model

struggles due to low-light conditions, partial animal visibility (e.g., only a leg is visible), or high similarity between species. However, our two-stage approach successfully addresses

these issues. By leveraging specialized expert models, it accurately identifies the species even under these challenging conditions, demonstrating its robustness and effectiveness. RIGHTS

AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in

any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The

images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is

not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission

directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE

Mulero-Pázmány, M., Hurtado, S., Barba-González, C. _et al._ Addressing significant challenges for animal detection in camera trap images: a novel deep learning-based approach. _Sci Rep_ 15,

16191 (2025). https://doi.org/10.1038/s41598-025-90249-z Download citation * Received: 29 October 2024 * Accepted: 11 February 2025 * Published: 09 May 2025 * DOI:

https://doi.org/10.1038/s41598-025-90249-z SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a shareable link is not

currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative KEYWORDS * Camera traps * Deep learning * Animal identification *

YOLO