RGB-D cameras have been commercialized, and many applications using them have been proposed. In this paper, we propose a robust registration method of multiple RGB-D cameras. We use a human body tracking system provided by Azure Kinect SDK to estimate a coarse global registration between cameras. As this coarse global registration has some error, we refine it using feature matching. However, the matched feature pairs include mismatches, hindering good performance. Therefore, we propose a registration refinement procedure that removes these mismatches and uses the global registration. In an experiment, the ratio of inliers among the matched features is greater than 95% for all tested feature matchers. Thus, we experimentally confirm that mismatches can be eliminated via the proposed method even in difficult situations and that a more precise global registration of RGB-D cameras can be obtained.

Multi-object tracking is a significant field in computer vision since it provides essential information for video surveillance and analysis. Several different deep learning-based approaches have been developed to improve the performance of multi-object tracking by applying the most accurate and efficient combinations of object detection models and appearance embedding extraction models. However, two-stage methods show a low inference speed since the embedding extraction can only be performed at the end of the object detection. To alleviate this problem, single-shot methods, which simultaneously perform object detection and embedding extraction, have been developed and have drastically improved the inference speed. However, there is a trade-off between accuracy and efficiency. Therefore, this study proposes an enhanced single-shot multi-object tracking system that displays improved accuracy while maintaining a high inference speed. With a strong feature extraction and fusion, the object detection of our model achieves an AP score of 69.93% on the UA-DETRAC dataset and outperforms previous state-of-the-art methods, such as FairMOT and JDE. Based on the improved object detection performance, our multi-object tracking system achieves a MOTA score of 68.5% and a PR-MOTA score of 24.5% on the same dataset, also surpassing the previous state-of-the-art trackers.

Computer-based data processing capabilities have evolved to handle a lot of information. As such, the complexity of three-dimensional (3D) models (e.g., animations or real-time voxels) containing large volumes of information has increased exponentially. This rapid increase in complexity has led to problems with recording and transmission. In this study, we propose a method of efficiently managing and compressing animation information stored in the 3D point-clouds sequence. A compressed point-cloud is created by reconfiguring the points based on their voxels. Compared with the original point-cloud, noise caused by errors is removed, and a preprocessing procedure that achieves high performance in a redundant processing algorithm is proposed. The results of experiments and rendering demonstrate an average file-size reduction of 40% using the proposed algorithm. Moreover, 13% of the over-lap data are extracted and removed, and the file size is further reduced.

People counting in surveillance cameras is a key technology for understanding the flow population and generating heat maps. In recent years, people detection performance has been greatly improved with the development of object detection algorithms using deep learning. However, in places where people are crowded, the detection rate is low as people are often occluded by other people. We proposed a people-counting method using a stereo camera to resolve the non-detection problem due to the occlusion. We applied stereo matching to extract the depth image and convert the camera view to top view using depth information. People were detected using a height map and an occupancy map, and people were tracked and counted using a Kalman filter-based tracker. We operated the proposed method on the NVIDIA Jetson TX2 to check the real-time operation possibility on the embedded board. Experimental results showed that the proposed method had higher accuracy than the existing methods and that real-time processing is possible.