The aim of this project will be to improve upon a Deep Learning model that extracts morphological features from microscopy images of Bacillaria Paradoxa. This will allow our organization to begin creating a database for the study of movement behavior and non-neuronal cognition in simple multicellular organisms. You will be improving upon the Digital Bacillaria project, which was started in the Summer of 2019. This year our main aim is to enhance the existing deep learning model (implemented in TensorFlow), as well as integrate the model into our species-specific library of machine learning models (DevoWorm AI). You will be involved in pre-processing and analyzing microscopy videos from our database of Bacillaria movement, along with tweaking the model for greater generalization. Integration of the model will involve adding functionality in the form of an interactive GUI, which will allow our community to analyze and display the data in terms of interesting behavioral variables. The successful applicant will be proficient in Python, C++, the basics of Machine Learning libraries and computer vision, HTML, and CSS.
DevoWormAI: link
Digital Bacillaria project: link
Raw video (microscopy) data link
Here are the tab-delimited versions of the data link
Paper with analysis from 2019 link
Recent presentation on Bacillaria movement link