Welcome to ConsNet API’s documentation!

ConsNet API is a part of the ConsNet project, it provides several useful functionalities and standard evaluation metrics for the task of human-object interaction (HOI) detection. Currently supported functionalities include but are not limited to:

• Data Loading

  • Load HICO-DET annotations into a tensor

  • Convert HICO-DET annotations to COCO format

• Data Processing

  • Convert IDs among actions, objects and interactions

  • Compute intersection-over-unions (IoUs) among human-object pairs

  • Perform non-maximum suppression (NMS) on human-object pairs

• Standard Evaluation

  • Evaluate HOI detection results on HICO-DET dataset

Getting Started

• Installation
• Usage

API Reference

• consnet.api.data
  • convert_anno()
  • load_anno()
  • obj_idx_to_coco_id()
  • hoi_idx_to_coco_id()
  • hoi_idx_to_act_idx()
  • hoi_idx_to_obj_idx()
  • act_idx_to_hoi_idx()
  • obj_idx_to_hoi_idx()
  • get_act_name()
  • get_obj_name()
  • get_act_and_obj_name()
  • get_hoi_name()
  • get_rare_hoi_idx()
  • get_non_rare_hoi_idx()
  • get_seen_hoi_idx()
  • get_unseen_hoi_idx()
  • scatter_hum_to_hoi()
  • scatter_act_to_hoi()
  • scatter_obj_to_hoi()
• consnet.api.bbox
  • pair_iou()
  • pair_nms()
• consnet.api.evaluation
  • hico_det_eval()

Citation

If you find this project useful for your research, please kindly cite our paper.

@inproceedings{liu2020consnet,
  title={ConsNet: Learning Consistency Graph for Zero-Shot Human-Object Interaction Detection},
  author={Liu, Ye and Yuan, Junsong and Chen, Chang Wen},
  booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
  pages={4235--4243},
  year={2020}
}