My main research directions include multimodal perception and recognition, light field imaging and processing, computer vision in industrial scenarios, as well as research in transfer learning, domain adaptation, industrial applications of vision foundation models, and integrated perception-recognition-decision-control systems.

Multimodal perception and recognition is a research field that combines information from different types of sensors or data sources to enhance perception and recognition capabilities. These data sources can include images, videos, audio, LiDAR, sensor data, etc. The research in this field aims to achieve a more comprehensive and accurate understanding of the environment or object recognition results by integrating data from multiple modalities.

Light field imaging and processing is a research direction that involves the acquisition, analysis, and application of light field data. Light field imaging technology can record information on the direction and intensity of light rays, capturing multi-view information of every point in a scene. Unlike traditional 2D images, light field images contain rich three-dimensional spatial information, making them widely applicable in various fields.

Computer vision in industrial scenarios applies computer vision technology to industrial environments to achieve tasks such as automated inspection, condition recognition, quality control, and robot guidance. This field combines computer science, artificial intelligence, image processing, and machine learning technologies to improve production efficiency, reduce costs, and ensure product quality.

Transfer learning and domain adaptation are two important methods in machine learning and deep learning used to improve model performance on new domains or tasks. They aim to leverage existing data and model knowledge to reduce the need for large amounts of labeled data in new domains, thereby improving training efficiency and model performance.

Integrated perception-recognition-decision-control systems is a research direction that integrates sensing technology, recognition technology, and control systems to achieve full-process automation and intelligence from environment perception, information recognition to decision control. This technology is widely used in industrial automation, intelligent manufacturing, autonomous driving, and other fields.