Multi-view Tracking, Re-ID, and Social Network Analysis of a Flock of Visually Similar Birds in an Outdoor Aviary

Xiao, Shiting; Wang, Yufu; Perkes, Ammon; Pfrommer, Bernd; Schmidt, Marc F.; Daniilidis, Kostas; Badger, Marc

doi:10.1007/s11263-023-01768-z

Cited by 7 publications

(5 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: Discussionmentioning

confidence: 99%

“…However, this data acquisition method is generally time-consuming and labor-intensive. With the development of computer vision and deep learning technologies, object detection techniques hold the potential to provide a low-cost and efficient means of data acquisition for bird surveys, thus enabling researchers to conveniently access key information on bird population dynamics, migration patterns, and habitat conditions[50][51][52].However, the application of object detection technology in bird surveys and monitoring research faces challenges due to the lack of available large-scale bird object detection datasets and limited technical references for evaluating detection model performance and model lightweighting performance in bird detection tasks. To address these challenges, we conducted in-depth research on bird object detection dataset construction, detection model performance comparisons, and model lightweighting.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Bird Object Detection: Dataset Construction, Model Performance Evaluation, and Model Lightweighting

Wang,

Zhou,

Zhang

et al. 2023

Animals

View full text Add to dashboard Cite

The application of object detection technology has a positive auxiliary role in advancing the intelligence of bird recognition and enhancing the convenience of bird field surveys. However, challenges arise due to the absence of dedicated bird datasets and evaluation benchmarks. To address this, we have not only constructed the largest known bird object detection dataset, but also compared the performances of eight mainstream detection models on bird object detection tasks and proposed feasible approaches for model lightweighting in bird object detection. Our constructed bird detection dataset of GBDD1433-2023, includes 1433 globally common bird species and 148,000 manually annotated bird images. Based on this dataset, two-stage detection models like Faster R-CNN and Cascade R-CNN demonstrated superior performances, achieving a Mean Average Precision (mAP) of 73.7% compared to one-stage models. In addition, compared to one-stage object detection models, two-stage object detection models have a stronger robustness to variations in foreground image scaling and background interference in bird images. On bird counting tasks, the accuracy ranged between 60.8% to 77.2% for up to five birds in an image, but this decreased sharply beyond that count, suggesting limitations of object detection models in multi-bird counting tasks. Finally, we proposed an adaptive localization distillation method for one-stage lightweight object detection models that are suitable for offline deployment, which improved the performance of the relevant models. Overall, our work furnishes an enriched dataset and practice guidelines for selecting suitable bird detection models.

show abstract

Section: Discussionmentioning

confidence: 99%

mentioning

confidence: 99%

Bird Object Detection: Dataset Construction, Model Performance Evaluation, and Model Lightweighting

Wang,

Zhou,

Zhang

et al. 2023

Animals

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

“…Automatic posture estimation is not only less labor-intensive but also achieves greater objectivity than manual behavior coding based on ethograms [61]. Classification of posture-based measurements can provide direct information about individual and group behaviors such as social and dominance interactions [60], decision making and information transfer [47,62] and courtship behaviour [50,63]. Marker-based systems have been recently employed to measure the 3D head orientation of birds, presenting an effective method for estimating visual field use, lateralization of eye use, and attentional foci of birds [35,62,[64][65][66].…”

Section: Introductionmentioning

confidence: 99%

“…The lack of adoption of 3D animal posture tracking can be attributed to a few main challenges. Firstly, algorithms for multiple individuals are less mature due to correspondence matching (identifying and matching the same points across different frames or views) and object trajectory tracking [60]. Secondly, 3D tracking often requires multiple synchronized camera views, which is either expensive or difficult to set up outside of the lab [50].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Peering into the world of wild passerines with 3D-SOCS: synchronized video capture and posture estimation

Chimento,

Chan,

Aplin

et al. 2024

Preprint

View full text Add to dashboard Cite

Collection of large behavioral data-sets on wild animals in natural habitats is vital in ecology and evolution studies. Recent progress in machine learning and computer vision, combined with inexpensive microcomputers, have unlocked a new frontier of fine-scale markerless measurements. Here, we leverage these advancements to develop a 3D Synchronized Outdoor Camera System (3D-SOCS): an inexpensive, mobile and automated method for collecting behavioral data on wild animals using synchronized video frames from Raspberry Pi controlled cameras. Accuracy tests demonstrate 3D-SOCS’ markerless tracking can estimate postures with a 3mm tolerance. To illustrate its research potential, we place 3D-SOCS in the field and conduct a stimulus presentation experiment. We estimate 3D postures and trajectories for multiple individuals of different bird species, and use this data to characterize the visual field configuration of wild great tits (Parus major), a model species in behavioral ecology. We find their optic axes at approximately±60° azimuth and −5° elevation. Furthermore, birds exhibit individual differences in lateralization. We also show that birds’ convex hulls predicts body weight, highlighting 3D-SOCS’ potential for non-invasive population monitoring. In summary, 3D-SOCS is a first-of-its-kind camera system for wild research, presenting exciting potential to measure fine-scaled behaviour and morphology in wild birds.

show abstract

Irregular Frame Rate Synchronization of Multi-camera Videos for Data-Driven Animal Behavior Detection

Dovdon,

Agarwal,

Dajsuren

et al. 2024

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Multi-view Tracking, Re-ID, and Social Network Analysis of a Flock of Visually Similar Birds in an Outdoor Aviary

Cited by 7 publications

References 76 publications

Bird Object Detection: Dataset Construction, Model Performance Evaluation, and Model Lightweighting

Bird Object Detection: Dataset Construction, Model Performance Evaluation, and Model Lightweighting

Peering into the world of wild passerines with 3D-SOCS: synchronized video capture and posture estimation

Irregular Frame Rate Synchronization of Multi-camera Videos for Data-Driven Animal Behavior Detection

Contact Info

Product

Resources

About