Asking For Help With the Traffic Accident Detection Model
Dear Enos,
We hope you are doing well.
We are Traffic.Ai, a capstone team at the University of Maryland, working on a project focused on detecting roadway crashes and assessing their severity and conditions.
We came across your model on Hugging Face and have been using it to detect vehicles and accidents from live traffic camera streams (via .m3u8 URLs). We’ve been very impressed with the model’s accuracy. It performs quite well for both vehicle and accident detection. However, we are currently facing challenges with processing live video streams. The model runs pretty slowly and experiences noticeable lag, even when running on higher-performance computers.
For context, we’ve attached a screenshot of how we are using the code in VS Code and a short screen recording of the current performance.
We were hoping you might be able to offer any guidance or suggestions on how to improve the processing speed and overall smoothness of the model. Any insights you could share would be extremely helpful to our project. Please let us know if you could help us with this.
We would be happy to acknowledge your contribution in our final project submission.
Thank you for your time and consideration.
