ferstar/dsh-tool-ocr

README 摘要

dsh-tool-ocr English 中文 A standalone DeepSeek Harness plugin: local image text recognition for models without vision input (e.g. DeepSeek chat models), backed by the standalone newbee-ocr ( nbocr ) engine over PP-OCRv6 models. Fully out-of-tree: depends only on published dsh base packages ( @deepseek-ai/cordis , @deepseek-ai/dsh-tools , @deepseek-ai/dsh-subprocess , …). No coupling to the deepseek-harness repository. What it provides Item Description ocr tool Model-facing tool: recognize / status / install (alias of check) / check actions Image inputs Local path (resolved against the session workspace) or attachment id (image already attached to the conversation, materialized to a temp file) Output Reading-ordered , engine facts, per-block bounding boxes ( include boxes ), review flags (low-confidence / amount / numeric / date / quantity, needsReview ), heuristic Markdown table ( table ) Robustness Survives MNN diagnostics polluting engine stdout; honors caller cancellation and timeout Two ways to use it Depending on which dsh build you run, image input works differently: Option A — official dsh: pass image paths ( ocr { path } ) The official deepseek-harness rejects image uploads …

分类