pranit-gandhi/detect-car-part

Car Parts Detection Parquet

The parquet file: detect-carparts.parquet comprises of carpart detections in the YouTube video: Should You Buy a 2026 Toyota RAV4? A Thorough Review by a Mechanic.

The parquet file: retrieve-carparts.parquet comprises of retrievals against the given query images in aegean-ai/rav4-exterior-images.

This was done as part of Assignment-2, titled 'Image-to-Video Semantic Retreival via Object Detection.'

---

Schema for detect-carparts

Attribute	Data Type	Description
video_id	string	Video ID (YcvECxtXoxQ)
frame_index	integer (64 bit)	1 FPS frame index (equivalent to timestamp in seconds)
class_label	string	Predicted class label (car part name)
bounding_box	array[float (64 bit)]	[x_min, y_min, x_max, y_max]
confidence_score	float (64 bit)	Detector confidence for the detection

---

Schema for retrieve-carparts

Attribute	Data Type	Description
qid	integer (64 bit)	Query ID (uniquely identifies each query)
class_label	string	Predicted class label (car part name)
start_timestamp	integer (64 bit)	Timestamp of the start of a particular carpart detection
end_timestamp	integer (64 bit)	Timestamp of the end of a particular carpart detection
number_of_supporting_detections	integer (64 bit)	Total number of detections in a particular segment (start to end)

---

Class ID to Label Mapping

0: back_bumper
1: back_door
2: back_glass
3: back_left_door
4: back_left_light
5: back_light
6: back_right_door
7: back_right_light
8: front_bumper
9: front_door
10: front_glass
11: front_left_door
12: front_left_light
13: front_light
14: front_right_door
15: front_right_light
16: hood
17: left_mirror
18: object
19: right_mirror
20: tailgate
21: trunk
22: wheel

---