datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
yolo26l_car_parts
Video Detections (YOLO26L Car Parts)
Dataset Schema
The data is stored in Apache Parquet format with snappy compression.
Column
Type
Description
video_id
string
YouTube ID for the source video.
frame_index
int64
The index of the frame in the video stream.
timestamp
float64
The exact time in the video where the detection occurred (in seconds).
class_label
string
The name of the car part (using Ultralytic's car-seg dataset).
x_min, y_min
float32
Top-left… See the full description on the dataset page: https://huggingface.co/datasets/Sliverwall/yolo26l_car_parts.yolo26l_car_parts_queried
Multi-Component Image-to-Video Mapping (YOLO26L)
Dataset Schema
The data is stored in Apache Parquet format with snappy compression. Each row represents a continuous video clip that matches a specific query image.
Column
Type
Description
instance_id
int64
Unique ID for the specific continuous sighting in the video.
source_image_index
int64
The index of the original query image from the reference aegean dataset.
components_queried
string
The set of car parts… See the full description on the dataset page: https://huggingface.co/datasets/Sliverwall/yolo26l_car_parts_queried.