Your question is Tagging Objects in Video. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How would you tag/label the various objects appearing in a video using multiple independent vision-recognition systems?
Asked in the technical_phone_screen_round1 stage. It followed a discussion of the candidate's experience, in a 45-minute round with no formal coding; the candidate felt it barely counted as a coding interview.
Explain a practical end-to-end design. Address how recognizers produce outputs, how detections are associated across frames, how conflicting labels are fused, and how results are stored and validated. Discuss batch versus online processing, confidence thresholds, unknown objects, computational cost, and failure handling.