相关搜索

Job Responsibilities Design and develop an automated annotation system to support tens-of-millions-scale robotics video datasets. Leverage spatiotemporal video features, scene-change detection, and multimodal semantic understanding to achieve precise segmentation of video clip boundaries. Combine VLMs (Vision-Language Models) with domain-specific models to...

Job Responsibilities Lead the development of the multimodal data system for embodied large models, covering data types such as images, video, language, interleaved image-text, action trajectories, and robot interaction data, to support model training, evaluation, and iteration. Design and build large-scale data processing and production pipelines, including...