Operationalizing Convolutional Neural Network Architectures for Prohibited Object Detection in X-Ray Imagery
This work addresses security screening automation for aviation and related fields, but it is incremental as it applies existing methods to a specific domain with optimizations.
The paper tackled prohibited object detection in X-ray imagery by evaluating CNN architectures like FreeAnchor with ResNet50, achieving mAP scores of 87.7 and 85.8 on benchmark datasets and an inference speed of ~13 fps, while showing resilience to image compression with only a 1.1% mAP decrease at JPEG level 50.
The recent advancement in deep Convolutional Neural Network (CNN) has brought insight into the automation of X-ray security screening for aviation security and beyond. Here, we explore the viability of two recent end-to-end object detection CNN architectures, Cascade R-CNN and FreeAnchor, for prohibited item detection by balancing processing time and the impact of image data compression from an operational viewpoint. Overall, we achieve maximal detection performance using a FreeAnchor architecture with a ResNet50 backbone, obtaining mean Average Precision (mAP) of 87.7 and 85.8 for using the OPIXray and SIXray benchmark datasets, showing superior performance over prior work on both. With fewer parameters and less training time, FreeAnchor achieves the highest detection inference speed of ~13 fps (3.9 ms per image). Furthermore, we evaluate the impact of lossy image compression upon detector performance. The CNN models display substantial resilience to the lossy compression, resulting in only a 1.1% decrease in mAP at the JPEG compression level of 50. Additionally, a thorough evaluation of data augmentation techniques is provided, including adaptions of MixUp and CutMix strategy as well as other standard transformations, further improving the detection accuracy.