Liang Wang

Liang Wang

Auto-generated profile, based on publicly available information

Liang Wang

School of Mechatronics Engineering and Automation, Shanghai University, Shanghai, China

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 October 2019 in IEEE Access
Verified
Journal Article
Open access
YXYulin Xu
Yulin Xu
LWLiang Wang
Liang Wang
AYAolei Yang
Aolei Yang
CLChen Ling
Chen Ling
This paper proposes GraspCNN, an approach to grasp detection where a feasible robotic grasp is detected as an oriented diameter circle in RGB image, using a single convolutional neural network. By detecting robotic grasps as oriented diameter circles, grasp representation is thereby simplified. In addition to our novel grasp representation, a grasp pose localization algorithm is proposed to project an oriented diameter circle back to a 6D grasp pose in point cloud. GraspCNN predicts feasible grasping circles and grasp probabilities directly from RGB image. Experiments show that GraspCNN achieves a 96.5% accuracy on the Cornell Grasping Dataset, outperforming existing one-stage detectors for grasp detection. GraspCNN is fast and stable, which can process RGB image at 50 fps and meet the requirements of real-time applications. To detect objects and locate feasible grasps simultaneously, GraspCNN is executed in parallel with YOLO, which achieves outstanding performance on both object detection and grasp detection.

1 August 2012
Verified
Conference Paper
Closed access
WLWang Liang
Wang Liang
JSJunfang Song
Junfang Song
At present the method of detecting traffic video speed is the research hotspot and difficulty of world intelligent transportation (ITS) field. It obtains the instantaneous speed of the vehicle through the corresponding video image processing algorithms. Based on the working mechanism of inductive loop detectors and radar (microwave) detector, this paper introduces the concept of virtual loops of video, and proposes a new virtual loops of video speed detection method. In this method, firstly obtain actual distance of all the pixels by manually setting the relevant points in first frame, then identify and track rear after the vehicles on the virtual loops, get the rear position of vehicles linked to several frame, finally calculate the speed of the vehicle according to the rear position. Simulation results show that this method can derive the instantaneous speed accurately, its measurement precision reaches 95% in actual application, satisfies the requirements of traffic supervision department, and verifies the availability and rationality.

1 September 2013
Verified
Conference Paper
Closed access
JHJun Hu
Jun Hu
ZYZhenlong Yuan
Zhenlong Yuan
GZGuofu Zhu
Guofu Zhu
LWLiang Wang
Liang Wang
XHXiaotao Huang
Xiaotao Huang
Constant false alarm rate (CFAR) methods are generally applied for the target detection after clutter mitigation in the through-wall sensing applications using ultra-wide band through-wall radar. Current studies of CFAR only focus on detection of moving human targets without considering the effect of residual clutters. However, in the cases with the presence of strong stationary clutters, little residual clutters after clutter mitigation are strong enough to cause false alarms in the following target detection, compared with the weak human target echoes. Unfortunately, conventional CFAR methods fail to remove the effect of residual clutters and suffer from false alarms. To combat this problem and obtain a reliable detection performance, a moving target CFAR detection method along slow-time profile is proposed in this paper. Experiments with an impulse through wall radar show that the proposed method not only has good detection performance for moving human targets, but also successfully gets rid of the problems brought by the strong residual stationary clutters.