Weakly Supervised 2D Pose Adaptation and Body Part Segmentation for Concealed Object Detection

Abstract

Weakly supervised pose estimation can be used to assist unsupervised body part segmentation and concealed item detection. The accuracy of pose estimation is essential for precise body part segmentation and accurate concealed item detection. In this paper, we show how poses obtained from an RGB pretrained 2D pose detector can be modified for the backscatter image domain. The 2D poses are refined using RANSAC bundle adjustment to minimize the projection loss in 3D. Furthermore, we show how 2D poses can be optimized using a newly proposed 3D-to-2D pose correction network weakly supervised with pose prior regularizers and multi-view pose and posture consistency losses. The optimized 2D poses are used to segment human body parts. We then train a body-part-aware anomaly detection network to detect foreign (concealed threat) objects on segmented body parts. Our work is applied to the TSA passenger screening dataset containing millimeter wave scan images of airport travelers annotated with only binary labels that indicate whether a foreign object is concealed on a body part. Our proposed approach significantly improves the detection accuracy of TSA 2D backscatter images in existing works with a state-of-the-art performance of 97% F1-score, 0.0559 log-loss on the TSA-PSD test-set, and a 74% reduction in 2D pose error.

Visual Parking Occupancy Detection Using Extended Contextual Image Information via a Multi-Branch Output ConvNeXt Network

“…This sub-block combines the information across channels, following an inverted bottleneck design [ 30 ], in which the hidden dimension is four times wider than the input and output of the sub-block. The architecture design was presented by MobileNetV2 [ 36 ], and since then it has gained popularity in several advanced architectures [ 37 , 38 ].…”

Computer Vision in Human Analysis: From Face and Body to Clothes

“…In particular, the accuracy is improved for motions related to humans and objects. Amadi et al [ 2 ] analyze the segmentation of human body parts through the usage of optimized 2D poses, validating the approach on the Transportation Security Administration Passenger Screening Dataset (TSA-PSD). The task of 3D human pose estimation is addressed in [ 3 ], in which the authors propose the use of bidirectional gated recurrent units to predict the global motion sequence from the local pose sequence.…”