Vision AI Project Submission Template
Use Case Title
(E.g., “Helmet Detection on Construction Sites”)
Team Members
List all names here.
(E.g., Ayesha Khan, Rahul Patel, Neha Sharma, Arjun Rao)
Use Case Importance (2 marks)
Briefly explain why this problem matters (2 sentences).
(E.g., Ensuring safety compliance on construction sites can reduce injuries and legal risk. Our solution helps detect missing helmets in real time.)
Data Collection and Annotation (2 marks)
- How did you collect the data?
- What classes were annotated?
- Which annotation tool did you use?
- How many total images?
- Include 2-3 sample images.
(E.g., We captured 500 images using a phone camera in different lighting. We used Makesense/Roboflow/labelme to annotate two classes: “Helmet” and “No Helmet.”)
Model Training and Validation (2 marks)
- What model and version (e.g., YOLOvX-Nano) did you train?
- What metrics did you monitor (e.g., mAP, precision, recall)?
- Did you collect more data after initial training? Why?
(E.g., We trained YOLOv9 with 400 training and 100 validation images. After low precision on “No Helmet,” we added 150 new samples and retrained.)
Model Deployment and Demo Video (7 marks)
- How did the model perform in the real world (FPS, accuracy)?
- How many models you deployed in the YOLOvX mobile app?
- Record a demo screen video (30secs to 1min) and put Google drive link here (before LinkedIn post).
Conclusion (2 marks)
Summarize your results and any challenges or learnings.
(E.g., Our model achieved 85% accuracy in live tests. We learned that consistent lighting greatly affects performance.)
Here are 15 marks and other 35 marks will be derived from the test.