ComfyUI Node
RapidOCR Text Mask (PP-OCR)
Runs PP-OCR models through RapidOCR ONNXRuntime and turns every text detection into a standard ComfyUI MASK. CUDA requires onnxruntime-gpu.
RapidOCR Text Mask (PP-OCR)
- image
- mask
- detections_json
◄speed_profilebalanced►
◄minimum_confidence0.72►
◄padding8►
◄mask_shapepolygon►
◄acceleratorauto►
◄cpu_threads0►
◄invert_maskfalse►
CategoryDeepSeek OCR
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| speed_profile | COMBO | balanced | 3 options: fast, balanced, thorough |
| minimum_confidence | FLOAT | 0.720–1 | — |
| padding | INT | 80–256 | — |
| mask_shape | COMBO | polygon | 2 options: polygon, rectangle |
| accelerator | COMBO | auto | 3 options: auto, cpu, cuda |
| cpu_threads | INT | 00–128 | — |
| invert_mask | BOOLEAN | false | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| mask | MASK | — |
| detections_json | STRING | — |