Assignment 2 – Image-to-Video Semantic Retrieval (Car Exterior Parts)
Artifacts
detections.parquet: frame-level detections for the retrieval corpus video.
intervals.parquet: contiguous presence intervals derived from detections.
retrieval_results.parquet: batch retrieval outputs for all query images (top-k labels, top-n intervals).
retrieval_summary.parquet: compact per-query summary (best interval per predicted label).
Source: YouTube video… See the full description on the dataset page:
https://huggingface.co/datasets/maazimam24/assignment2-car-retrieval.
