Video violence detection has gained significant attention in recent years due to its applications in surveillance and security. This paper proposes a two-stage framework for detecting violent actions in video sequences. The first stage leverages GMFlow, a pre-trained optical flow network, to capture the temporal motion between consecutive frames, effectively encoding motion dynamics. In the second stage, we integrate these optical flow images with RGB frames and feed them into a CBAM-enhanced ResNet3D network to capture complementary spatiotemporal features. The attention mechanism provided by CBAM enables the network to focus on the most relevant regions in the frames, improving the detection of violent actions. We evaluate the proposed framework on three widely used datasets: Hockey Fight, Crowd Violence, and UBI-Fight. Our experimental results demonstrate superior performance compared to several state-of-the-art methods, achieving AUC scores of 0.963 on UBI-Fight and accuracies of 97.5% and 94.0% on Hockey Fight and Crowd Violence, respectively. The proposed approach effectively combines GMFlow-generated optical flow with deep 3D convolutional networks, providing robust and efficient detection of violence in videos.

Table detection in document images is a challenging problem due to diverse layouts, irregular structures, and embedded graphical elements. In this study, we present HTTD (Hierarchical Transformer for Table Detection), a cutting-edge model that combines a Swin-L Transformer backbone with advanced Transformer-based mechanisms to achieve superior performance. HTTD addresses three key challenges: handling diverse document layouts, including historical and modern structures; improving computational efficiency and training convergence; and demonstrating adaptability to non-standard tasks like medical imaging and receipt key detection. Evaluated on benchmark datasets, HTTD achieves state-of-the-art results, with precision rates of 96.98% on ICDAR-2019 cTDaR, 96.43% on TNCR, and 93.14% on TabRecSet. These results validate its effectiveness and efficiency, paving the way for advanced document analysis and data digitization tasks.

Understanding information extracted from receipts is a critical task for real-world applications such as financial tracking, auditing, and enterprise resource management. In this paper, we introduce ReceiptQA, a novel large-scale dataset designed for receipt understanding through question-answering (QA). ReceiptQA contains 171,000 question–answer pairs derived from 3500 receipt images, constructed via two complementary methodologies: (1) LLM-Generated Dataset: 70,000 synthetically generated QA pairs, where each receipt is paired with 20 unique, context-specific questions. These questions are produced using a state-of-the-art large language model (LLM) and validated through human annotation to ensure accuracy, relevance, and diversity. (2) Human-Created Dataset: 101,000 manually crafted questions spanning answerable and unanswerable queries. This subset includes carefully designed templates of varying difficulty (easy/hard) to comprehensively evaluate QA systems across diverse receipt domains. To benchmark performance, we evaluate leading vision–language models (VLMs) and language models (LMs), including GPT-4o, Phi-3B, Phi-3.5B, LLaVA-7B, InternVL2 (4B/8B), LLaMA-3.2, and Gemini. We further fine-tune a LLaMA-3.2 11B model on ReceiptQA, achieving significant improvements over baseline models on validation and test sets. Our analysis uncovers critical strengths and limitations of existing models in handling receipt-based QA tasks, establishing a robust benchmark for future research.

Colorization in X-ray material discrimination is considered one of the main phases in X-ray baggage inspection systems for detecting contraband and hazardous materials by displaying different materials with specific colors. The substructure of material discrimination identifies materials based on their atomic number. However, the images are checked and assigned by a human factor, which may decelerate the verification process. Therefore, researchers used computer vision and machine learning methods to expedite the examination process and ascertain the precise identification of materials and elements. This study proposes a color-based material discrimination method for single-energy X-ray images based on the dual-energy colorization. We use a convolutional neural network to discriminate materials into several classes, such as organic, non-organic substances, and metals. It highlights the details of the objects, including occluded objects, compared to commonly used segmentation methods, which do not show the details of the objects. We trained and tested our model on three popular X-ray datasets, which are Korean datasets comprising three kinds of scanners: (Rapiscan, Smith, Astrophysics), SIXray, and COMPASS-XP. The results showed that the proposed method achieved high performance in X-ray colorization in terms of peak-signal-to-noise ratio (PSNR), structural similarity index (SSIM), and learned perceptual image patch similarity (LPIPS). We applied the trained models to the single-energy X-ray images and we compared the results obtained from each model.

This study addresses the field of X-ray security screening and focuses on synthesising realistic X-ray images using advanced generative models. Insufficient training data in this area pose a major challenge, which we address through innovative data augmentation techniques. We utilise the power of generative adversarial networks (GANs) and conditional GANs (cGANs), in particular the Pix2Pix and Pix2PixHD models, to investigate the generation of X-ray images from various inputs such as masks and edges. Our experiments conducted on a Korean dataset containing dangerous objects relevant to security screening show the effectiveness of these models in improving the quality and realism of image synthesis. Quantitative evaluations based on metrics such as PSNR, SSIM, LPIPS, FID, and FSIM, with scores of 19.93, 0.71, 0.12, 29.36, and 0.54, respectively, show the superiority of our strategy, especially when integrated with hybrid inputs containing both edges and masks. Overall, our results highlight the potential of advanced generative models to overcome the challenges of data scarcity in X-ray security screening and pave the way for more efficient and accurate inspection systems.

Light field (LF) technology has become a focus of great interest (due to its use in many applications), especially since the introduction of the consumer LF camera, which facilitated the acquisition of dense LF images. Obtaining densely sampled LF images is costly due to the trade-off between spatial and angular resolutions. Accordingly, in this research, we suggest a learning-based solution to this challenging problem, reconstructing dense, high-quality LF images. Instead of training our model with several images of the same scene, we used raw LF images (lenslet images). The raw LF format enables the encoding of several images of the same scene into one image. Consequently, it helps the network to understand and simulate the relationship between different images, resulting in higher quality images. We divided our model into two successive modules: LFR and LF augmentation (LFA). Each module is represented using a convolutional neural network-based residual network (CNN). We trained our network to lessen the absolute error between the novel and reference views. Experimental findings on real-world datasets show that our suggested method has excellent performance and superiority over state-of-the-art approaches.