This repository contains a structured semantic index of a car exterior video, generated using an object detection model fine-tuned on the Ultralytics Car Parts Segmentation dataset.
The purpose of this dataset is to enable scalable image-to-video semantic retrieval by separating:
Video detection (offline processing)
Retrieval (query-time lookup)
The Parquet file serves as the semantic index of the video.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/sharushanK/car-parts.
