I am building a project that can classify students in a real world classroom as attentive or inattentive and provide a report to teacher, and i am planning to include one exam mode, which takes the snapshot of suspicious activity and send to teacher, I am an absolute beginner I have 3-4 months time in hand, can you help me by giving methadology and roadmap to start
@GK-I I recently made a similar comment on another post, but I’d say question number one is ‘where am I going to find my dataset of relevant labeled training images (in this case, attentive/inattentive students)’ ?
No data— no model.
So start there.
@GK-I I will grant you though there may be another way to solve your project problem-- And that is to use real-time face tracking via a library like OpenCV.
In that case your dataset could be developed by building a dataset of attentive/inattentive actions.
However these are a completely different set of algorithms, so we’re no longer talking YOLO/CNNs as you stated you were looking for.
1 Like