2D-Pose Based Human Body Segmentation for Weakly-Supervised Concealed Object Detection in Backscatter Millimeter-Wave Images.
Lawrence AmadiGady AgamPublished in: ICPR Workshops (1) (2022)
Keyphrases
- human body
- weakly supervised
- pose estimation
- object detection
- segmentation algorithm
- object detectors
- object class
- segmentation method
- three dimensional
- superpixels
- object recognition
- image segmentation
- test images
- articulated objects
- object segmentation
- bounding box
- partial occlusion
- body parts
- image features
- human motion
- feature points
- image retrieval
- computer vision
- image data
- input image
- lighting conditions
- object categories
- level set
- face detection
- image sequences
- topic models
- multiple images
- multiple objects
- multiscale