In this paper, we propose a novel framework for multi-target multi-camera tracking (MTMCT) of vehicles based on metadata-aided re-identification (MA-ReID) and the trajectory-based camera link model (TCLM). Given a video sequence and the corresponding frame-by-frame vehicle detections, we first address the isolated tracklets issue from single camera tracking (SCT) by the proposed traffic-aware single-camera tracking (TSCT). Then, after automatically constructing the TCLM, we solve MTMCT by the MA-ReID. The TCLM is generated from camera topological configuration to obtain the spatial and temporal information to improve the performance of MTMCT by reducing the candidate search of ReID. We also use the temporal attention model to create more discriminative embeddings of trajectories from each camera to achieve robust distance measures for vehicle ReID. Moreover, we train a metadata classifier for MTMCT to obtain the metadata feature, which is concatenated with the temporal attention based embeddings. Finally, the TCLM and hierarchical clustering are jointly applied for global ID assignment. The proposed method is evaluated on the CityFlow dataset, achieving IDF1 76.77%, which outperforms the state-of-the-art MTMCT methods.

The recent research by deep learning has shown many breakthroughs with high performance that were not achieved with traditional machine learning algorithms. Particularly in the field of object detection, commercial products with high accuracy in real environment are applied through the deep learning methods. However, the object detection method using convolutional neural network(CNN) has a disadvantage that a large number of feature maps should be generated in order to be robust against scale change and occlusion of the object. Also, simply raising number of feature maps does not improve performance. In this paper, we propose to integrate additional prediction layers into conventional Yolo-v3 using spatial pyramid pooling to complement the detection accuracy of the vehicle for large scale changes or being occluded by other objects. Our proposed detector achieves 85.29% mAP, which outperformed than those of the DPM, ACF, R-CNN, CompACT, NANO, EB, GP-FRCNN, SA-FRCNN, Faster-R CNN2, HAVD, and SSD-VDIG on the UA-DETRAC benchmark data-set consisting of challenging real-world-traffic videos.

Gait recognition is one of technology for biometrics at a distance that can be used to identify a human through walking postures and body shape. In the field of information forensics and security, gait recognition is exploited for crime prevention, forensic identification, and social security. However, the existing gait recognition methods usually consider the appearance, posture and temporal information separately, thus, we focus on considering a learned temporal attention mechanism to fuse these features in global and local manner. In this paper, a novel gait recognition framework Temporal Attention and Keypoint-guided Embedding (GaitTAKE) is proposed, which effectively fuses temporal-attention-based global and local appearance feature and temporal aggregated human pose feature. Experimental results show that our proposed method achieves superior performance in gait recognition with rank-1 accuracy of 98.0% (normal), 97.5% (bag) and 92.2% (coat) on the CASIA-B gait dataset; 90.4% accuracy on the OU-MVLP gait dataset; 51.34% accuracy on the GREW dataset; 53.1% accuracy on the Gait3D gait dataset.

As the powerful performance of deep learning has been proven, many computer vision researchers have applied deep learning methods to their works as a breakthrough that could not be achieved with conventional computer vision algorithms. Particularly in pathological image analysis, deep learning plays an important role because some diagnosis requires a considerable cost or much time. In a recent, convolutional neural network (CNN)-based deep learning models have shown meaningful results in pathological image analysis, reducing time and cost. However, existing CNN-based segmentation models perform the same convolution operation for all channels of a feature map. It could be an inefficient operation according to information theory. We propose (Shannon) entropy-based convolutional module (ECM) for efficient convolutional operation in terms of a communication system. The fundamental coding manner of a communication system based on information theory is to allocate fewer bits for data showing the high probability of occurrence, and vice versa. Following up this coding manner, a feature is divided into dominant and recessive features according to the channel importance calculated from the channel attention module, and a heavy operation is conducted on the recessive feature and a light operation is conducted on the dominant feature. This operating manner can make a network perform efficient calculations and improve its performance. Furthermore, our proposed module is a portable unit, thus it can be a replacement of any convolution without modification of the whole architecture. To the best of our knowledge, our proposed module is the first trial to mimic the coding manner of information theory. The models equipped with our proposed module outperform the original models achieving 0.855 of F1 score and 0.832 of Jaccard score on colorectal cancer (CRC) image data-set.