This file is the detection corpus for Assignment 2: image-to-video semantic retrieval.
File: video_detections_50_epochs.parquet
Rows: 1100 (one row per detection)
Source video: YcvECxtXoxQ
Frame sampling interval: 5 seconds
Detector: YOLO segmentation model fine-tuned for 50 epochs
video_id
string
YouTube video id for the corpus video.
frame_index
int
Frame number from sampled frames… See the full description on the dataset page:
https://huggingface.co/datasets/ksam98nyu/assignment-2-rav4-semantic-search.
