H2FA R-CNN: Holistic and Hierarchical Feature Alignment for Cross-domain Weakly Supervised Object Detection.
Yunqiu XuYifan SunZongxin YangJiaxu MiaoYi YangPublished in: CVPR (2022)
Keyphrases
- cross domain
- weakly supervised
- object detection
- object class
- object detectors
- knowledge transfer
- transfer learning
- object categories
- topic models
- text categorization
- relation extraction
- object recognition
- face detection
- image features
- feature vectors
- multi class
- weak classifiers
- superpixels
- object segmentation
- object classes
- named entities
- computer vision
- unsupervised learning
- semi supervised
- bounding box
- feature selection