Evaluation dataset of 5 rendered kitchen scenes with per-frame bounding-box annotations, staged in a cluttered home kitchen, to measure a YOLOv8 object detection model's accuracy on partially occluded kitchen utensils for home service robots.
This dataset mirrors public data-pack render outputs from Physicl.
Each row represents one render view. The image column contains a stable URL to the primary render image uploaded under /data;… See the full description on the dataset page:
https://huggingface.co/datasets/physicl-test/pack-ohj-of-my-project-co3gkd-a6718212.
