Research Highlights on Computer Vision

Scaling Visual Language Models for Remote Sensing with Open Data and LLMs

Introduction Vision-language models (VLMs) show great promise for interpreting remote sensing (RS) imagery in human-like ways. However, training effective VLMs requires massive amounts of paired image-text data - a significant challenge in the RS domain.

Last updated on Jun 3, 2025

Scaling Visual Language Models for Remote Sensing with Open Data and LLMs

Intelligent diagnosis of DCD

We developed digital visual-motor systems and algorithms to autonomously evaluate developmental coordination disorder (DCD) and delve into associated pathologies.

Last updated on Dec 27, 2023

Temporal action detection

We developed deep learning methods to automatically discover action segments from the untrimmed videos.

Last updated on Dec 22, 2023