Haichang Li

Synesthesia: Music visualization for the hearing impaired

Have you ever thought about synesthesia? There are about 200 million hearing impaired people in the world, and the music we normally enjoy is a luxury for them. For example, when we hear grand piano music, we may conjure up images of medieval battlefields, but those who are equally imaginative and hard of hearing cannot. So, our project is willing to be their ears, based on emotions and AIGC and LLM, we let them use the "see" way to "hear" the music. In the face of the injustices they suffer, our answer takes its own approach, using technology to provide them with a potentially viable approach.

Social Robot for the Depressed and Lonely

To help people with special needs, such as those with depression or the elderly who are alone, cope with mental health challenges, we propose an innovative approach that uses multimodal social robots, combined with sentiment analysis and natural language interaction technologies.

The reason for using multimodality is to improve accessibility and accuracy, with multimodal data verifying each other, while also taking into account that some people may not be able to output through a certain mode (such as situations with limited facial expressions, typing difficulties, or inability to speak). The application of sentiment analysis, which helps us understand users' emotions, and LLM, which provide a more human conversational experience through natural language processing, play a key role in this approach. Through this integrated, multimodal approach, our social robots are able to more fully understand and support the mental health needs of our target population, contributing to their emotional well-being.

ShineResume: Multimodal AIGC platform for confused graduates

The Post-COVID-19 era has brought a cold winter to the Chinese job market, with companies announcing layoffs and hiring freezes. In such an environment, there exists a group of underserved individuals who come from ordinary schools, and during their college years, they simply follow the prescribed curriculum. They don't know what they're passionate about, what kind of work suits them, or how to start their job search. They find themselves trapped in a cocoon of information gap.

We aim to have AI collaborate with these individuals, putting people at the center, leveraging multimodal AIGC and LLM. With just the user's background information, we can recommend various industries and positions, extract key terms, and then create and optimize their resumes as per requirements. Additionally, we can perform style transfer for their official photos to meet the desired appearance.

Our purpose is to let AI be valueable for underserved people, bring benefits to humanity, and help neglected people have the opportunity to find themselves and shine.

Data driven model for human thermal confort during sports

In order to help architects design the interior environment, the interaction between the human body's psychological subjective thermal comfort and the external objective environment is worth studying and observing. Subjective feelings can be reflected through the designed user study, and the PMV model and a variety of data analysis and processing methods can be combined to quantify and predict such psychological feelings, so as to improve the interaction experience between people's psychological feelings and the external physical environment.