Korean VLM GUI Agent Data | 10,000 Sets | Large Language Model(LLM) Data

This dataset contains 10,000 high-quality GUI interaction sequences (continuous frames) across mobile, tablet, and PC devices, covering scenarios such as desktop application operations, web browsing behaviors, and more. Each sequence includes detailed annotations of user actions, UI element states, task objectives, and execution results, along with complete operational logic chains to enhance agents' interface comprehension and task automation capabilities. All data is anonymized and complies with global privacy regulations including GDPR, CCPA, and PIPL, ensuring full legal and ethical adherence in collection and usage. Date size 10,000 sets Image resolution matches device resolution App types pre-installed apps, shopping, food ordering, entertainment, education, etc. Devices mobile, tablet, PC Data diversity multiple apps, instructions, and operations Annotations instructions, fine-grained actions, COT, AI assistant labels, operation areas Data format .jpg, .png, .json