HOI detection is a fundamental problem in computer vision as it provides semantic information about the interactions among the detected objects. The current HOI benchmarks, e.g., HCIO-Det and V-COCO, aim to tackle Many kinds of relationships between human and objects which results in imbalance data distribution and the lack of practical significance.
Aug. 22th, 00:05:00, 2019
Aug. 31th, 00:05:00, 2019
Oct. 18th, 00:05:00, 2019
Oct. 20th, 00:05:00, 2019
Oct. 28th, 2019
Training / validation set released
Testing set released and submission opened
Challenge winners notified
Winners present at ICCV 2019 Workshop
- Make sure the submission format has met the requirements. See results.json
- Using extra training data are not allowded. But the usage of pretrain model has no limitations.
- You have 10 submission chances in total.
- The evaluation process can takes times. And a failed submission will not cause the reduction of submission chances.
- Following the standard settings in HICO-DET and VCOCO benchmarks, we evaluate HOI detection using mean average precision (mAP). An HOI detection is considered as a true positive when the human detection, the object detection, and the interaction class are all correct. The human and object bounding boxes are considered as true positives if they overlap with a ground truth bounding boxes of the same class with an intersection over union (IoU) greater than 0.5.