I am now a 2-year master student of the Communication and Multimedia Laboratory affiliated at the Department of CSIE, National Taiwan University, supervised by Prof. Yung-Yu Chuang. Before that, I received my B.Eng degree in Automation from Nanjing Normal University in 2017. My research interests include Computer Vision (e.g., Person re-ID) and Computational Photography. I am fortunate to work closely with Dr. Zheng Wang and Prof. Yinqiang Zheng.
The hand gesture recognition plays an important role in the human-computer interaction (HCI) field. Previous works mainly focused on researching this task in the shorter distance, which can only be applied in a limited scenario due to the recognition distance constrains. In this paper, we propose a method, which is named Joint Single Shot Multibox Detector (JSSD) network, to solve the hard long-distance hand gesture recognition task. The method is based on the framework of SSD. Our model employed two SSDs for object detection. One is for the head-shoulder region detection and the other is for the hand gesture bounding box proposal and classification. We validate our proposed model on real data collected from USB camera at several distance levels. The experiment results show that our method is more robust and precise in recognizing the hand gestures at a longer distance. Concretely, JSSD network is able to recognize the hand gestures captured up to 6 meters away from the camera.
This paper proposes an approach that uses a parallel array to improve the reliability and robustness of logical stochastic resonance subject to colored noise. The experimental results demonstrate that (i) the increase of array size can extend the optimal range of noise intensity and increase the maximum probability of obtaining correct logic operation. (ii) The optimal range of noise correlation time is broadened with the increase of array size. (iii) The main difference between logical stochastic resonance subject to white noise and colored noise is that the increase of noise correlation time broadens the optimal range of noise intensity when the stochastic noise is colored noise. At the same time, the maximum probability of obtaining correct logic operation is close to 1. Therefore, reliable and robust logic gate can be realized under a certain array size.
Tags: #Computer Vision, #Image-based Retrieval
Search cross Night and Day: TBA
Feasible Polarimetric Camera Calibration: TBA.
Polarimetric Camera Self-calibration : TBA.
Logical Stochastical Resonance: We proposed two effective methods to improve the performance of Logic Stochastic Resonance based on observation of system characteristics. One uses a parallel array to interact with noise correlation time and the other utilizes the asymmetry of the nonlinear system. I published one first author paper (SCI) and also won the Excellent Bachelor Thesis Award.
Autonomous Tree Climbing Robot: We used a Kinect to capture the 3D point cloud of the tree. Then, we generate a feasible and optimal motion and trajectory based on the specified start and end point. It is obtained according to the 3D tree model combined with the previously built climbing stability constraints. Lastly, we developed the embedded control algorithm making the robot move autonomously according to the planned result.
``Freescale Cup'' Intelligent Car Race: We developed two car-like mobile robots, one using electromagnetic sensors and the other using CMOS camera, as well as a wheeled inverted pendulum robot which can track complex road autonomously with high speed (≥ 2.5m/s). My efforts involve signal processing, robot position estimation, control algorithm developing and tuning.
Jan. 2018 - present
Details: Polarimetric Imaging and Heterogeneous Person Re-identification.
Jul. 2016 - Aug. 2016
Details: Autonomous Tree Climbing Robot. Worked at Prof. Yangsheng Xu's Lab.
Details: Rigid body geometric interpolation.