Yiyang Zhang

Bio: Yiyang Zhang is an academic researcher from Beihang University. The author has contributed to research in topics: Engineering & Sketch recognition. The author has an hindex of 1, co-authored 1 publications receiving 9 citations.

Design concept of sign language recognition translation and gesture recognition control system based on deep learning and machine vision

Abstract: With the development of society, gestures are used in many aspects, but the computer's functionality for gesture recognition is still to be improved. This article is mainly a preliminary idea of a basic gesture recognition system built based on the existing Google deep learning framework TensorFlow and gesture recognition components in MediaPipe and OpenCv machine vision open-source library. The training dataset is first subjected to skeleton key point coordinate extraction, then the pre-processed dataset is used to train the neural network and constitute the preliminary model, and finally the model is corrected and changed in the end.

Performance comparative of OpenCV Template Matching method on Jetson TX2 and Jetson Nano developer kits

Abstract: Template Matching is a widely used method for object detection in digital images, it requires great processing power since it is an exhaustive method that compares the intensity levels of a source image pixel-to-pixel with a template image that contains the object to identify. Nowadays there are dedicated embedded systems that provide high processing capabilities, such as the NVIDIA Jetson family. This work presents the experimentation and the performance comparison between the Jetson Nano and Jetson TX2 development kits, when implementing the Template Matching method, in order to get an evaluation criterion to select one of them in image processing projects. It was carried out to six images with different sizes and two variants in terms of the size of the template image. The processing times for the sequential implementation using the CPUs and the parallel implementation with the GPUs were obtained quantitatively. It was observed that the processing times using the parallel versions on average doubled those of the sequential versions and that the Jetson TX2 exceeded the Jetson Nano in execution speeds.

Water Target Recognition Method and Application for Unmanned Surface Vessels

Abstract: Water target recognition is a critical challenge for the perception technology of unmanned surface vessels (USVs). In the application of USV, detection accuracy and the inference time both matter, while it is tough to strike a balance and single-frame water target detection behaves unstable in the video detection. To solve these problems, many strategies are applied to increase YOLOv4’s performance, including network pruning, the focal loss function, blank label training, and preprocessing with histogram normalization. The optimized detection method achieves a mean average precision (mAP) of 81.74% and a prediction speed of 26.77 frames per second (FPS), which meets the USV navigation requirements. To build the integrated USV-based system for water target recognition, a water target dataset containing 9936 images is created from offshore USV experiments in which the human-in-the-loop annotation and mosaic data augmentation methods are used. The issues of miss detection and false alarm can be considerably mitigated by cascading the Siamese-RPN tracking network, and the major color of a water target can be retrieved using a local contrast saliency color detection scheme. The system being tested is called “ME120” includes an embedded edge computing platform (Nvidia Jetson AGX Xavier). Finally, online dataset learning demonstrates the improved YOLOv4 achieves an increase of 66.98% in FPS at the cost of a decrease of 0.79% in mAP when compared with the original YOLOv4 and offline navigation experiments validate that our system achieves high recognition capability while maintaining a high degree of robustness.

Water Target Recognition Method and Application for Unmanned Surface Vessels

Abstract: Water target recognition is a critical challenge for the perception technology of unmanned surface vessels (USVs). In the application of USV, detection accuracy and the inference time both matter, while it is tough to strike a balance and single-frame water target detection behaves unstable in the video detection. To solve these problems, many strategies are applied to increase YOLOv4’s performance, including network pruning, the focal loss function, blank label training, and preprocessing with histogram normalization. The optimized detection method achieves a mean average precision (mAP) of 81.74% and a prediction speed of 26.77 frames per second (FPS), which meets the USV navigation requirements. To build the integrated USV-based system for water target recognition, a water target dataset containing 9936 images is created from offshore USV experiments in which the human-in-the-loop annotation and mosaic data augmentation methods are used. The issues of miss detection and false alarm can be considerably mitigated by cascading the Siamese-RPN tracking network, and the major color of a water target can be retrieved using a local contrast saliency color detection scheme. The system being tested is called “ME120” includes an embedded edge computing platform (Nvidia Jetson AGX Xavier). Finally, online dataset learning demonstrates the improved YOLOv4 achieves an increase of 66.98% in FPS at the cost of a decrease of 0.79% in mAP when compared with the original YOLOv4 and offline navigation experiments validate that our system achieves high recognition capability while maintaining a high degree of robustness.