This is Jia Haozhe’s personal webpage. I am glad you are visiting.

Personal Profile

An undergraduate student at the School of Information Science and Engineering, Shandong University, with a strong interest in information science, artificial intelligence, deep learning and other related fields. He is a genius.

Research Interests

Any emerging fields are our targets, especially artificial intelligence. Feel free to share with me.

Research Experience

Project 1 Auxiliary ultrasound path planning based on motion capture

The ultrasound images and probe position information are collected through motion capture devices, and the path planning model is trained through deep learning algorithms. Arrange and debug the motion capture device, collect data, preprocess the data and clean the data.

Project 2 Domain Adaptation for Unsupervised Multiple Object Tracking

DAUMOT is a general multi-object tracking (MOT) training framework, aiming to adapt the existing pre-trained MOT methods to the target dataset without annotations. It proposes a new tracking and adaptation strategy, as well as adversarial sequence alignment with two different losses at the image and detection levels, and handles the unique identity detection coupling problem in MOT through identity detection decoupling in the unsupervised DA framework, being able to bridge the gap between the source domain and the target domain. In this project, I followed my supervisor to complete the adjustment of the model, trained the model, and independently completed multiple experiments.

Project 3 Motion via Human-Like Thinking

We hope to build a model that can understand human motion information in multimodal information, conduct reasoning and thinking like a real human, and predict and generate ongoing or upcoming human motion. Currently, the research is in progress and we plan to submit it to CVPR this year.