As a prototype for the AI logic behind a mobile application that can be used to detect different types of snakes, we trained different Convolutional Neural Network (CNN) models from the Tensorflow 2 Detection Model Zoo in an effort to benchmark different methods, build a model that works for our purposes, and of course, to learn.
Download the Pre-processed Snake Images dataset from Kaggle, and follow the instructions on training a custom object detector with Tensorflow 2. We have provided annotations in XML format for a number of images from the dataset (500 from the training set, 100 from the testing set) which can be found in the annotations/ directory.
To submit changes to this repo, please do the following steps:
- Make your changes in a new branch
- Upload your ipynb code as a .py so we can compare changes
- Open a Pull Request (PR) in Github, attempting to merge your branch into main, and request reviews from the group
- When you have approvals, merge your pull request
- Daniel Siegel - 101367445
- Michael McAllister - 101359469
- Hom Kandel - 101385341
- Eduardo Bastos de Moraes - 101345799
- Juan Clackworthy - 101372229
- Leandra Lai - 101367265