With the rapid development of CNN or Transformer, the present mainstream approaches regard an image patch as the reference of the target to perform tracking, which is known as template matching-based trackers. However, most existing template matching-based trackers only consider the per-frame localization accuracy, neglecting the potential distractor (similar object) dependencies among multiple video frames, which poses a fundamental challenge in template matching-based tracking. In this work, we propose a novel comprehensive framework with multi-frame distractor suppression for visual object tracking (MFDSTrack), which explicitly models the temporal history of both the target object and potential distractors. Specifically, we utilize a universal target candidate generation module to detect target candidates (both target and distractors), providing a holistic view of the scene. In addition, a temporal and distractor-aware association module is designed to suppress multi-frame distractors by adopting a simple encoder-decoder Transformer architecture. The encoder accepts inputs of target candidates’ history, while the decoder takes current target candidate queries and the output of the encoder as inputs to associate current target candidate queries with historical trajectories. We extensively evaluate our trackers, MFDSTrack-SD, MFDSTrack-OS, MFDSTrack-GRM, and MFDSTrack-LT on the LaSOT, LaSOText, TrackingNet, GOT-10k, UAV123, NFS, and OTB100 benchmark. Extensive experiments show that our methods outperform previous state-of-the-art trackers on seven tracking benchmarks.

The accurate breeding of individual sheep has shown outstanding effectiveness in food quality tracing, prevention of fake insurance claims, etc., for which sheep identification is the key to guaranteeing its high performance. As a promising solution, sheep identification based on sheep face detection has shown potential effectiveness in recent studies. Unfortunately, the performance of sheep face detection has still been a challenge due to diverse background illumination, sheep face angles and scales, etc. In this paper, an effective and lightweight sheep face detection method based on an improved RetinaFace algorithm is proposed. In order to achieve an accurate and real-time detection of sheep faces on actual sheep farms, the original RetinaFace algorithm is improved in two main aspects. Firstly, to accelerate the speed of multi-scale sheep face feature extraction, an improved MobileNetV3-large with a switchable atrous convolution is optimally used as the backbone network of the proposed algorithm. Secondly, the channel and spatial attention modules are added into the original detector module to highlight important facial features of the sheep. This helps obtain more discriminative sheep face features to mitigate against the challenges of diverse face angles and scale in sheep. The experimental results on our collected real-world scenarios have shown that the proposed method outperforms others with an F₁score of 95.25%, an average precision of 96.00%, a model size of 13.20 M, an average processing time of 26.83 ms, and a parameter of 3.20 M.

Multi-modal image matching is very challenging due to the significant diversities in visual appearance of different modal images. Typically, the existing well-performed methods mainly focus on learning invariant and discriminative features for measuring the relation between multi-modal image pairs. However, these methods often take the features as a whole and largely overlook the fact that different scale features for a same image pair may have different similarity, which may lead to sub-optimal results only. In this work, we propose a Scale-Separative Metric Learning Quadruplet network (SSML-QNet) for multi-modal image patch matching. Specifically, SSML-QNet can extract both relevant and irrelevant features of imaging modality with the proposed quadruplet network architecture. Then, the proposed Scale-Separative Metric Learning module separately encodes the similarity of different scale features with the pyramid structure. And for each scale, cross-modal consistent features are extracted and measured by coordinate and channel-wise attention sequentially. This makes our network robust to appearance divergence caused by different imaging mechanism. Experiments on the benchmark dataset (VIS-NIR, VIS-LWIR, Optical-SAR, and Brown) have verified that the proposed SSML-QNet is able to outperform other state-of-the-art methods. Furthermore, the cross-dataset transferring experiments on these four datasets also have shown that the proposed method has powerful ability of cross-dataset transferring.

The prediction of seismic intensity in an accurate and timely manner is needed to provide scientific guidance for disaster relief. Traditional seismic intensity prediction methods rely on seismograph equipment, which is limited by slow response times and high equipment costs. In this study, we introduce a low-computational-cost transformer-based (LCCTV) visual tracking method to predict seismic intensity in surveillance videos. To this end, an earthquake video dataset is proposed. It is captured in the laboratory environment, where the seismic level is obtained through seismic station simulation. With the proposed dataset, a low-computational-cost transformer-based visual tracking method is first proposed to estimate the movement trajectory of the calibration board target in videos in real time. In order to further improve the recognition accuracy, we then utilize a Butterworth filter to smooth the generated movement trajectory so as to remove low-frequency interference signals. Finally, the seismic intensity is predicted based on the velocity and acceleration derived from the smoothed movement trajectory. Experimental results demonstrated that the LCCTV outperformed other state-of-the-art approaches. The findings confirm that the proposed LCCTV can serve as a low-cost, scalable solution for seismic intensity analysis.

Change detection (CD) is a crucial task in remote sensing image analysis, including several fundamental tasks such as binary change detection (BCD), semantic change detection (SCD), etc. Recently, deep learning-based models (e.g., Convolutional neural networks (CNN) and Transformers) have made impressive progress in the field of remote sensing CD. However, those CD models are carefully designed for a specific fundamental task (BCD or SCD) and cannot achieve impressive progress in both BCD and SCD tasks simultaneously. In this paper, we designed a generalized architecture for both BCD and SCD tasks by investigating the common characteristics of these two tasks. In the feature extraction stage, a Multi-scale Attention Enhanced Encoder (MSAE) is introduced to extract global context and capture fine-grained features, which is beneficial for both BCD and SCD. In the training stage, a Change Detection Contrastive Learning Module (CDCL) is proposed to optimize feature distribution, improving the discriminative ability in distinguishing change regions and categories. To mitigate class imbalance issues in both BCD and SCD, we introduce Dynamic Rare-Aware Sampling for Change Detection (DRAS-CD), which dynamically prioritizes rare categories and enhances model robustness. Additionally, we collect the Yellow River Basin Semantic Change Detection dataset (YRSCD), which includes 13 change categories across diverse scenes with broad spatial and temporal coverage. Extensive experimental results on four public datasets (WHU-CD, LEVIR-CD, SECOND, Landsat-SCD) and YRSCD have shown superior performance in comparison to other state-of-the-art approaches, especially those based on the visual fundamental model, offering a generalized solution for both BCD and SCD tasks.