/nnc

https://github.com/devoworm/Proposals-Public-Lectures/blob/master/GSoC/2021/DL-for-Diatom-Non-neuronal-Cog.md

Primary LanguageJupyter Notebook

Diatom-Non-neuronal-Cognition

The aim of this project will be to improve upon a Deep Learning model that extracts morphological features from microscopy images of Bacillaria Paradoxa.

Notebooks:

Run in Google Colab - Object detection Run in Google Colab - Instance segmentation


Run from github

Setup:

  • Run sh setup_yolov5.sh for setting up and cloning yolov5 repository. OR
  • Run sh setup_scl_yolov4.sh for setting up and cloning Scaled yolov4 repository.

Training:

  • Make changes in train_yolov5.sh for training on dataset using yolov5.
    • To start training run sh train_yolov5.sh. OR
  • Make changes in train_scl_yolov4.sh for training on dataset using yolov4.
    • To start training run sh train_scl_yolov4.sh.

Prediction:

  • Make changes in predict_yolov5.sh for predicting on dataset using yolov5.
    • To make predicitions run sh predict_yolov5.sh. OR
  • Make changes in predict_scl_yolov4.sh for predicting on dataset using yolov4.
    • To make predictions run sh predict_scl_yolov4.sh.