giobbu / zebrafish

1-week project for Champalimaud Research

COLLECTIVE MOTION IN ZEBRAFISHES

Data

The data have been obtained using idtracker.ai. The data are related to trajectories for 60, 80 and 100 fish. Each subfolder has as a name the number of individuals in the trajectories.

• Data can be downloaded here

Zebrafish data analysis:

prob_simil:

• Fishes probability density within tank
• Cosine similarity among fishes to assess how parallel are all fish to each other

ML model to predict fish trajectory and speed

ml_model:

• LSTM/GRU for multi-step ahead fish speed and trajectory prediction

Focal fish ID-0

Focal fish ID-20

References

This repository is based on the following works:

• idTracker: tracking individuals in a group by automatic identification of unmarked animals
• Deep attention networks reveal the rules of collective motion in zebrafish