论文信息 - idtracker.ai: Tracking all individuals in large collectives of unmarked animals

idtracker.ai: Tracking all individuals in large collectives of unmarked animals

Our understanding of collective animal behavior is limited by our ability to track each of the individuals. We describe an algorithm and software, idtracker.ai, that extracts from video all trajectories with correct identities at a high accuracy for collectives of up to 100 individuals. It uses two deep networks, one detecting when animals touch or cross and an-other one for animal identification, trained adaptively to conditions and difficulty of the video.

[1] Xi En Cheng,et al. Zebrafish tracking using convolutional neural networks , 2017, Scientific Reports.

[2] Mj Rasch,et al. Closing the loop: tracking and perturbing behaviour of individuals in a group in real-time , 2016, bioRxiv.

[3] Hanqing Zhang,et al. ToxId: an efficient algorithm to solve occlusions when tracking multiple animals , 2017, Scientific Reports.

[4] Nicole Rusk. Deep learning , 2015, Nature Methods.

[5] Yoshua Bengio,et al. Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[6] Sandra Martins,et al. Toward an Integrated Zebrafish Health Management Program Supporting Cancer and Neuroscience Research. , 2016, Zebrafish.

[7] Vicenç Quera,et al. A method for resolving occlusions when multitracking individuals in a shoal , 2015, Behavior research methods.

[8] A. Pérez-Escudero,et al. idTracker: tracking individuals in a group by automatic identification of unmarked animals , 2014, Nature Methods.

[9] Yan Qiu Chen,et al. Robust tracking of fish schools using CNN for head identification , 2017, Multimedia Tools and Applications.

[10] Jimmy Ba,et al. Adam: A Method for Stochastic Optimization , 2014, ICLR.

[11] Geoffrey E. Hinton,et al. Deep Learning , 2015, Nature.

[12] Guy Theraulaz,et al. Domino-like propagation of collective U-turns in fish schools , 2017, bioRxiv.