Object Detection

EventKitchen provides a total of 13,482 bounding boxes across 12 object classes: 1. Bowl, 2. Box, 3. Chopping board, 4. Cup, 5. Fork, 6. Knife, 7. Pan, 8. Plastic packaging, 9. Plate, 10. Spatula, 11. Spoon, and 12. Teapot. Two human annotators were first assigned to annotate the synchronized RGB data at a frequency of one frame every four seconds. The RGB annotations were then projected onto the synchronized event data streams using the calibration matrices and depth stream. More details of the annotation pipeline and statistics are reported in the paper.

As described in Download, annotations for the left event camera are saved in objects/cooking_activity_LeftEvent_bbox.csv in each recording-session tree, while annotations for the right event camera are saved in objects/cooking_activity_RightEvent_bbox.csv. We also provide the original RGB annotations in objects/cooking_activity_DRGB_bbox.csv. More details about the annotation format and a script for loading the annotations are available in the EventKitchen toolkit.