Skip to content

Latest commit

 

History

History
35 lines (19 loc) · 2.29 KB

README.md

File metadata and controls

35 lines (19 loc) · 2.29 KB

CuttleShuttle-Analysis

Analysis code for the Cuttle Shuttle video dataset. Requires Visual Studio Code (2012 or later) and Bonsai.

Project website: http://www.everymind.online/CuttleShuttle/

Video dataset

CuttleShuttle-VideoDataset-Raw

Video annotator

Open the Visual Studio solution "Bonsai.VideoAnnotations.sln" found in /Bonsai.VideoAnnotations/. Press F5 or hit the Start button. A new instance of Bonsai will open. From inside this instance, open the Bonsai workflow "CuttleShuttle-VideoAnnotator.bonsai" found in /Workflows/ShaderNavigator/.

This workflow is used to hand-annotate the location of the center of the cuttlefish's mantle and the tip of its tail, then save a snapshot of the annotated frame.

To mark the center of the cuttlefish's mantle, move the mouse cursor until the tip of it points to the spot you believe to be the mantle center, then hit the "F" key. To mark the tail, move the mouse cursor until the tip of it points to the very tip of the cuttlefish's tail, then hit the "V" key. To save a snapshot of that frame, hit the "G" key. All keys noted here are in the QWERTY layout.

To ensure that the annotations are correctly saved into a .csv file, exit the VideoNavigator visualizer window (where you are annotating the video frames) before stopping the entire workflow.

When changing the video file you wish to annotate or view, change the filepath in the node labeled "String". To double check that the workflow made all of the correct updates, check all nodes labeled "VideoFile", "VideoNavigator", and "CsvReader" to make sure the file names are consistent.

This workflow can also be used to visualize previous annotations and add more annotations.

Training dataset for machine learning

Frame snapshots and annotations.

Potentially relevant and useful other projects

Deep Lab Cut: markerless tracking of user-defined features in videos of animal behaviour

https://github.com/AlexEMG/DeepLabCut

https://arxiv.org/abs/1804.03142v1