Speaker

Dr. Tian Qiu

Wuyi University, China

Speech Title: A liquid-lens based defect inspection algorithm for optical isolators

Abstract: As one of the basic optical components, optical isolator is widely used in optical communication equipment. Currently the inspection for optical isolator is mainly manually done by visual inspection by workers. The manual inspection method is time-consuming and laborious, and it often hurts  inspectors’ eyesight. Based on the physical structure of the optical isolator, an automatic defect inspection algorithm using liquid lens is proposed. The proposed algorithm can first locate the surface of the transparent optical isolator, then inspect several defects on the surface of the optical isolator. The experiments show that the algorithm achieves satisfactory results in the detection inspection, which can reduce the manpower demand of the visual inspection to less than 10%.

A. Prof. Antti Rasila

Technion – Israel Institute of Technology, Guangdong Technion

Speech Title: Conformal mappings in mathematical modeling

A. Prof. Wei Jiang

Zhejiang University, China

Speech Title: Research on Person Re-identification in Intelligent Visual Surveillance

Abstract: Person Re-identification (ReID) is a very important part of an intelligent video surveillance system, which aims at retrieving all videoes/images of a specific person from the candidate set. It can also achieve quick person retrieval from the large-scale video database captured by a surveillance camera network and be integrated into other related tasks, such as person historical trajectory analysis, person tracking, and motion analysis. However, those complicated factors, such as pose variations, occlusion,illumination changes, and camera viewpoint differences, bring challenges to achieve fast and accurate ReID. In this talk, we presented a novel method for  visible-infrared cross-modality person re-identification.Visible cameras cannot work under dark conditions, so intelligent video surveillance systems introduce infrared cameras, and need to process both visible and infrared images. To solve the visible-infrared cross-modality ReID problem, we proposed  a visible-infrared cross-modality ReID method based on cross-spectrum dual-subspace pairing by generating input images from multiple spectrum, forcing the network to discover shared cross-modality features existing in all spectrum. we also presented a visible-infrared cross-modality ReID method based on modality transfer and dual-level unified representation. At the image level, image information is fused, while at the feature level, a hierarchical granularity triplet loss is designed. On datasets with short-wave near-infrared or long-wave far-infrared images, both metods have achieved state-of-the-art performance.