Sparrow Vis: A framework to analyse and annotate sparrow provisioning videos

This repository contains all code used for the thesis: Testing the silver spoon effect in a passerine using novel deep learning and computer vision pipeline. Please read project structure below to access seperate directories for each part of the project.

Please request access to the full repository in the Zenodo repository to run ceratain parts of the project which may contain sensitive data. The Zenodo repository also contains additional data samples.

Language & Dependencies

Please see README in seperate directories for section specific dependencies.

Project Structure

Directories and its contents:

• DeepMeerkat: Code and guidance to run Deep Meerkat
• MaskRCNN: Code and guidance to run MaskRCNN and obtain masks of sparrows
• Analysis: Code to run all silver spoon analysis
• Writeup:: Contain figures and supplementary information for writeup
• Pipeline: Code to process videos from Deep Meerkat Output to events to clips
• GenerateTraining: Generate training data by matching meerkat events with previously annotated data
• A) DeepMeerkat Framework: Training classification models for Deep Meerkat frames
• B) LRCN Framework: Training classification models for 7 second clips
• C) MaskRCNN Framework: Trianing classification models for MaskRCNN masks

Deep Learning approaches used

Author and Affiliations

Alex Chan Hoi Hang
[email protected]
MRes Computational Methods in Ecology and Evolution
Department of Life Sciences
Imperial College Silwood Park
UK. SL5 7PY