Reading up on computer-vision

The paper proposes Qwen-Image-Agent, a unified agentic framework that uses planning, reasoning, search, memory, and feedback to bridge the context gap in real-world text-to-image generation.

Baidu releases Unlimited-OCR, a VLM-based OCR that processes entire documents in a single pass using reference sliding window attention to avoid memory overflow.

Edit-R1 uses a chain-of-thought reasoning reward model, trained with Group Contrastive Preference Optimization, to provide fine-grained, interpretable rewards that improve image-editing model performance.

An AI tool from Google DeepMind, tested by U.S. Ski and Snowboard athletes, revealed flaws in Olympic snowboarder Maddie Mastro's crippler trick.

A thread breaks down the difficulty of building an AI that can recognize computer interfaces from raw video and reason with them without APIs.