Zishan Zeng
Computer Science undergraduate at the Southern University of Science and Technology. My interests include real-time visual perception for robotics, edge deployment, and vision-language models for image understanding and retrieval.
Research interests
I am interested in efficient, robust, and deployable visual intelligence for real-world settings. My current focus includes robotics vision and real-time perception, vision-language models and multimodal retrieval, edge deep-learning deployment, and quality-efficiency analysis for generative models.
Selected experience
Multi-turn multimodal image retrieval
Contributing to FashionAM training, evaluation, and ablation studies on multi-turn context, intent switching, and vision-language alignment.
RoboMaster robotics vision
Served as vision group lead and advisor, responsible for Jetson deployment, on-robot debugging, and low-light optimization.
Vision-language models and deployment
Built a LLaVA-style image-captioning project and a Jetson Nano face-recognition and expression-interaction system.
Education and skills at a glance
- B.Eng. candidate in Computer Science and Technology, SUSTech, Sep. 2023 - Present.
- GPA 3.71/4.0, major rank 78/152. Computer Vision 97 (A+), Artificial Intelligence 94 (A), Design and Analysis of Algorithms 94 (A).
- Regularly use C++ and Python, with project experience in PyTorch, TensorRT, Transformers, Jetson, ROS, and Git.