pkurainbow / fcos

FCOS: Fully Convolutional One-Stage Object Detection

Hi there 👋

• 🔭 I am presently focusing on addressing fundamental problems in the area of AIGC. My work includes text-to-image generation/editing, text-to-video generation, training large language models (LLMs) using Megatron-LLM, and applying LLMs to visual content generation and comprehension. Additionally, I am working on unifying the understanding and generation/editing of image, video, and point-cloud data, with a particular emphasis on fundamental visual understanding tasks, such as semantic and panoptic segmentation, object detection, and visual generation tasks like image, video, and 3D-NeRF generation, at Microsoft Research Asia.