Video captioning aims to generate comprehensive and coherent descriptions of the video content, contributing to the advancement of both video understanding and generation. However, existing methods often suffer from motion-detail imbalance, as models tend to overemphasize one aspect while neglecting the other. This imbalance results in incomplete captions, which in turn leads to a lack of consistency in video understanding and generation. To address this issue, we propose solutions from two aspects: 1) Data aspect: We constructed the Harmonizing Motion-Detail 270K (HMD-270K) dataset through a two-stage pipeline: Motion-Detail Fusion (MDF) and Fine-Grained Examination (FGE). 2) Optimization aspect: We introduce the Caption Set Equivalence Reward (CSER) based on Group Relative Policy Optimization (GRPO). CSER enhances completeness and accuracy in capturing both motion and details through unit-to-set matching and bidirectional validation. Based on the HMD-270K supervised fine-tuning and GRPO post-training with CSER, we developed OwlCap, a powerful video captioning Multi-modal Large Language Model (MLLM) with motion-detail balance. Experimental results demonstrate that OwlCap achieves significant improvements compared to baseline models on two benchmarks: the detail-focused VDC (+4.2 Acc) and the motion-focused DREAM-1K (+4.6 F1).

Clone attack is considered as a severely destructive threat in Internet of Things (IoT), because: 1) the attack may be easily launched due to the deficiency of hardware architecture and the limited resources against physical capture and compromise and 2) it may trigger a large variety of insider and outsider attacks. Different from traditional clone attack detection approaches that ground on a large amount of data traversing the network (e.g., locations and identities), this article tackles this problem by answering the following fundamental questions: do we really need so much raw information? whether there is an alternative for local event detection by a node far away from that event? when acquiring/tracing global knowledge of a system/network, do we really need a global collection effort? These questions are of much importance in a large variety of networks. Specifically, this article provides a collective memory design for global topology and identity tracing (GTI Tracing), via a localized computing paradigm within neighborhood. This localized paradigm computationally builds a connection of identity and topology from time and space domain to a new computation domain. Such a computation domain retains four properties: 1) transitivity; 2) global convergence; 3) determinacy; and 4) causality. With byte-size information at an arbitrary device, it can recover and keep tracing global topology and identity information, and thus providing deterministic detection of clones. Both theoretical analysis and experimental study have shown the advantages of the proposed design in both detection accuracy and privacy protection, at a cost of light communication, storage, and computation overhead at each device.

In this paper, we propose a novel descriptor for face recognition on grayscale images, depth images and 2D+depth images. It is a compact and effective descriptor computed from the magnitude and the direction difference. It can be concatenated with conventional descriptors such as well-known Local Binary Pattern (LBP) and Weber Local Binary Pattern (WLBP), to enhance their discrimination capability. To evaluate the performance of our descriptor, we conducted extensive experiments on three types of images using four different databases. The experimental results demonstrate the robustness and superiority of our approach, and the performances of our new descriptor surpass that without magnitude and direction difference. At the end, we further compare our descriptor with Convolution Neural Network (CNN) to show the compactness and effectiveness of the proposed approach.