Disclaimer: This is a group capstone project. I didn't do all of the files that are here. The publication of this repository is for personal use of proof of work for job applications and for personal use of learning more about what my other teamates do in the project.
In this project, we have:
- Taken our own dataset, applied some augmentation methods presented in Roboflow. Added for related public dataset for variation.
- Made use of the YOLOv5, YOLOv8, Faster-RCNN, RT-DETR documentation for experimenting on the dataset.
- Created a pipeline for 2-sized inference for application.
- Created an interface for model and inference demonstration.
- Achieved the mAP50 of 0.861 for all devices in the IoT kit.
Here are some images for reference:
