DeepSeek-OCR 2: Visual Causal Flow
This addresses a bottleneck in image understanding for AI systems by introducing a novel cognitive-inspired method, though it is incremental in advancing existing encoder architectures.
The paper tackled the problem of rigid visual token processing in vision-language models by proposing DeepSeek-OCR 2 with DeepEncoder V2, which dynamically reorders tokens based on image semantics, achieving improved performance on complex layout images as evidenced by benchmark results.
We present DeepSeek-OCR 2 to investigate the feasibility of a novel encoder-DeepEncoder V2-capable of dynamically reordering visual tokens upon image semantics. Conventional vision-language models (VLMs) invariably process visual tokens in a rigid raster-scan order (top-left to bottom-right) with fixed positional encoding when fed into LLMs. However, this contradicts human visual perception, which follows flexible yet semantically coherent scanning patterns driven by inherent logical structures. Particularly for images with complex layouts, human vision exhibits causally-informed sequential processing. Inspired by this cognitive mechanism, DeepEncoder V2 is designed to endow the encoder with causal reasoning capabilities, enabling it to intelligently reorder visual tokens prior to LLM-based content interpretation. This work explores a novel paradigm: whether 2D image understanding can be effectively achieved through two-cascaded 1D causal reasoning structures, thereby offering a new architectural approach with the potential to achieve genuine 2D reasoning. Codes and model weights are publicly accessible at http://github.com/deepseek-ai/DeepSeek-OCR-2.