datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
clockface-synth-dataset-10k
clockface-synth
Rendered analog clocks with exact times. Training data for
clockface, a small model that reads a
clock face and returns the time.
The label is exact by construction. Blender is told where to put the hands,
so the time is known to arbitrary precision and there is no annotation step and
no annotation error. That property is the reason this project is tractable.
These are renders, not photographs. A score measured on this data is training
telemetry, never a result.… See the full description on the dataset page: https://huggingface.co/datasets/lyte-codes/clockface-synth-dataset-10k.dm-codes
DM codes dataset
The dataset contains photos of Data Matrix (DM) codes and their annotations. The photos were taken on an iPhone and annotated manually by a human.
The annotations contain text, which is encoded in the DM code and the pixel coordinates of the DM code vertices.
The vertices are: tl = top left, tr = top right, br = bottom right, bl = bottom left.
Attribute is_clean specifies whether the DM code on the image is expected to be easily readable. For every DM code, there is… See the full description on the dataset page: https://huggingface.co/datasets/shortery/dm-codes.ocr_samplecord-v2cub200-images-codes-embeddingscub200-images-codes