Zichen Gu
Zichen Gu has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Fadeng Wang
Infrastructure Inspection Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China
1 shared publication - Hao Wang
Infrastructure Inspection Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China
1 shared publication - Huchao Zhai
Guangdong Intercity Railway Operation Co., Ltd., Guangzhou 573100, China
1 shared publication - Le Wang
Infrastructure Inspection Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China
1 shared publication - Ning Wang
Infrastructure Inspection Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China
1 shared publication - Qiang Han
Infrastructure Inspection Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China
1 shared publication
Latest Publications
- Open Access
Cross-View Measurement of Adjacent Fastener Bolt Spacing in Railway Turnouts Using Dual DLP Sensors Without Overlapping Fields of View
To measure the cross-view spacing between adjacent fastener bolts in railway turnouts, this study develops a dual-DLP-sensor structured-light measurement system without overlapping fields of view. A bridge-type calibration device is used to rapidly update the extrinsic parameters of the two DLP sensors. In a unified coordinate frame, the system integrates two-dimensional region-of-interest candidate generation, local three-dimensional geometric fitting, cross-view pairing, and measurement validity assessment to output bolt-spacing results. Experiments were conducted on 23 pairs of adjacent bolts with 15 repeated measurements using two DLP sensors. Under normal conditions, the mean absolute error, root mean square error, and average standard deviation were 0.261 mm, 0.290 mm, and 0.062 mm, respectively. Compared with fixed extrinsic parameters without updating, the bridge-based extrinsic update reduced the mean absolute error from 1.500 mm to 0.261 mm. The results indicate that the proposed task-driven dual-DLP-sensor measurement system can achieve stable cross-view spacing measurement with explicit validity criteria under non-overlapping fields of view, repeated deployment, and varying on-site data quality.
Open AccessCross-View Measurement of Adjacent Fastener Bolt Spacing in Railway Turnouts Using Dual DLP Sensors Without Overlapping Fields of View
To measure the cross-view spacing between adjacent fastener bolts in railway turnouts, this study develops a dual-DLP-sensor structured-light measurement system without overlapping fields of view. A bridge-type calibration device is used to rapidly update the extrinsic parameters of the two DLP sensors. In a unified coordinate frame, the system integrates two-dimensional region-of-interest candidate generation, local three-dimensional geometric fitting, cross-view pairing, and measurement validity assessment to output bolt-spacing results. Experiments were conducted on 23 pairs of adjacent bolts with 15 repeated measurements using two DLP sensors. Under normal conditions, the mean absolute error, root mean square error, and average standard deviation were 0.261 mm, 0.290 mm, and 0.062 mm, respectively. Compared with fixed extrinsic parameters without updating, the bridge-based extrinsic update reduced the mean absolute error from 1.500 mm to 0.261 mm. The results indicate that the proposed task-driven dual-DLP-sensor measurement system can achieve stable cross-view spacing measurement with explicit validity criteria under non-overlapping fields of view, repeated deployment, and varying on-site data quality.
- Open Access
- 4 January 2022
- Published by MDPI AG in Mathematics
- Vol. 10 (1) , 139
Cascaded Cross-Layer Fusion Network for Pedestrian Detection
The detection method based on anchor-free not only reduces the training cost of object detection, but also avoids the imbalance problem caused by an excessive number of anchors. However, these methods only pay attention to the impact of the detection head on the detection performance, thus ignoring the impact of feature fusion on the detection performance. In this article, we take pedestrian detection as an example and propose a one-stage network Cascaded Cross-layer Fusion Network (CCFNet) based on anchor-free. It consists of Cascaded Cross-layer Fusion module (CCF) and novel detection head. Among them, CCF fully considers the distribution of high-level information and low-level information of feature maps under different stages in the network. First, the deep network is used to remove a large amount of noise in the shallow features, and finally, the high-level features are reused to obtain a more complete feature representation. Secondly, for the pedestrian detection task, a novel detection head is designed, which uses the global smooth map (GSMap) to provide global information for the center map to obtain a more accurate center map. Finally, we verified the feasibility of CCFNet on the Caltech and CityPersons datasets.
- 4 January 2022
- Published by MDPI AG in Mathematics
- Vol. 10 (1) , 139
Open AccessCascaded Cross-Layer Fusion Network for Pedestrian Detection
The detection method based on anchor-free not only reduces the training cost of object detection, but also avoids the imbalance problem caused by an excessive number of anchors. However, these methods only pay attention to the impact of the detection head on the detection performance, thus ignoring the impact of feature fusion on the detection performance. In this article, we take pedestrian detection as an example and propose a one-stage network Cascaded Cross-layer Fusion Network (CCFNet) based on anchor-free. It consists of Cascaded Cross-layer Fusion module (CCF) and novel detection head. Among them, CCF fully considers the distribution of high-level information and low-level information of feature maps under different stages in the network. First, the deep network is used to remove a large amount of noise in the shallow features, and finally, the high-level features are reused to obtain a more complete feature representation. Secondly, for the pedestrian detection task, a novel detection head is designed, which uses the global smooth map (GSMap) to provide global information for the center map to obtain a more accurate center map. Finally, we verified the feasibility of CCFNet on the Caltech and CityPersons datasets.