Positioning
PP-OCRv6_tiny_det is the ultra-lightweight text-detection model in the PP-OCRv6 family, intended for edge and IoT deployment. It has approximately 0.43 million parameters and a listed model size of 1.9 MB. It uses an LCNetV4 backbone and a RepLKFPN feature pyramid. (huggingface.co)
Capabilities and uses
- Locates text regions in images, including multilingual printed, handwritten, rotated, curved and artistic text. Its output can be passed to a separate recognition model to extract text. (huggingface.co)
- Suitable for OCR pipelines processing documents, scene images and other text-bearing images, particularly where deployment resources are constrained. (huggingface.co)
- PaddleOCR reports a detection Hmean of 80.6% on its internal multi-scenario evaluation set. This result is not directly comparable with PP-OCRv5/v4 figures measured on a different evaluation set. (paddleocr.ai)
Limitations and deployment
This is a detection component, not a standalone text-transcription model: full OCR requires a separate text-recognition component. Results depend on image quality, text and deployment settings. PaddleOCR provides local inference and training models; the official model card identifies the license as Apache 2.0. (huggingface.co)
Release date
The official PaddleOCR announcement dates the PaddleOCR 3.7.0 and PP-OCRv6 family release to June 11, 2026. It does not separately establish a distinct release date for this exact detector. (github.com)
Pricing
As of November 16, 2026, a model-specific hosted API price for PP-OCRv6_tiny_det could not be verified from official Baidu domestic or international sources. Separate input and output charges are therefore unverified. The downloadable model can be deployed locally, subject to infrastructure costs. Baidu publishes prices for a broader PP-OCRv6 OCR service, but its documentation does not establish that those prices apply to this specific standalone detector, so they are not presented as this model's price. (huggingface.co)