This repository implements the course project of CSE 521, AI & IoT for Medicine, Fall 2023.
We recommend a recent Python 3.7+ distribution of Anaconda with numpy
, torch
, and scikit-learn
installed.
See INSTALL
for detailed instructions on installation.
The implementation consists of two parts: synthetic image generator and object detector based on YOLOv5. Detailed guidance on how to run the code is available in HOW-TO-RUN
.
We use both a real-world dataset and a synthetic dataset to train our object detector. You can run our main.py
to create synthetic images together with the annotations associated with the objects.
Our object detector is based on YOLOv5. We provide a patch to the public YOLOv5 implementation under the folder yolo_patch
for our task-specific services, including the open-world object detection algorithm, a user-friendly color palette to visualize detection results, and an AI-assisted voice service.