datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
racing-gears
Racing Gear Digits
Image classification dataset for detecting gear numbers (0-9) from racing onboard camera telemetry overlays, supplemented with MNIST digits for robustness.
Similar in spirit to MNIST but for a specific real-world application: reading the gear indicator from racing car onboard video feeds in real-time.
Dataset
5,964 training / 1,003 validation images
32×32 grayscale PNG images
10 classes (digits 0-9)
Source column distinguishes racing-original… See the full description on the dataset page: https://huggingface.co/datasets/tobil/racing-gears.3d_gearsI created these images using a p5.js. I first rendered them in a GIF, uploaded to Google Drive, and then extracted the individual frames
from the GIF.
The light background gears images were generated in p5.js using this sketch.
The dark background gears images were generated in p5.js using this sketch.