Project · Artificial Intelligence (AI)

Ship a real-time defect detector

Train, quantise and deploy a detection model on a webcam feed with a latency budget. Includes the dataset, the deployment target and the rubric.

3 weeks · IntermediatePublished
Project preview

Preview available

Build brief preview

The brief is available below. Supporting repositories, datasets and project files are not available yet.

The brief

Week 1: dataset audit and baseline training. Week 2: quantisation and export, measure latency. Week 3: integrate with the camera loop, add a confidence threshold and a review queue.

What you submit: model card, latency table, a two-minute recording and the review-queue design.

Rubric emphasises the trade-off you chose between precision and latency, and how you justified it.