Shivdutta/S15-YOLOV9

Inference of Vehicle detection using Yolov9

• —This application showcases the inference capabilities of a Yolo v9 trained on the vehicle dataset from kaggle. Vehicle Dataset Repo Link
• —The model is trained on 6 classes:
• —car
• —threewheel
• —bus
• —truck
• —motorbike
• —van
• —The architecture is based on Yolo v9 papar https://arxiv.org/abs/2402.13616 and model is trained using https://github.com/WongKinYiu/yolov9.git
• —detect.py file used for inference.
• —From gradio applicaiton call is made to detect.py using command line shell with unique folder name passed as argument
• —After processing, image/video is picked from same location.

Mentioned below is the link for Training Repository Training Repo Link

• —Post training process, the model is saved locally and then uploaded to Gradio Spaces.
• —Attached below is the link to download model file

• —This app has two features :

• —Video Prediction: " - This feature will allow detection of moving vehicles in the the video
• —Image Prediction:
• —This feature will allow detection of vehicle in the the image

Usage:

• —Video Prediction: " - Upload video file and detect vehicles present in the video.
• —Inferencing is done using CPU therefore it takes more time.
• —Image Prediction:
• —Upload image file and detect vehicles present in the image.

[image]

[image]

[image]

[image]

Training repo:

https://github.com/Shivdutta/ERA2-Session15-Yolov9

Thank you