Nodes/ComfyUI-Grounding/Grounding Mask Detector
ComfyUI Node

Grounding Mask Detector

A ComfyUI node in grounding with 9 inputs and 3 outputs.

By PozzettiAndrea·Created 10 months ago·Updated 23 days ago· 45
Grounding Mask Detector
  • model
  • image
  • masks
  • overlaid_mask
  • text
promptSegment the main object in the image
confidence_threshold0.30
florence2_max_tokens1024
florence2_num_beams3
sa2va_max_tokens2048
sa2va_num_beams1
seed42
Categorygrounding

Inputs (9)

NameTypeDefaultDescription
modelMASK_MODEL
imageIMAGE
promptSTRINGSegment the main object in the imageFor Florence-2: descriptive phrase. For SA2VA: explicit segmentation instruction (e.g., 'Segment the person')
confidence_thresholdFLOAT0.300–1Confidence threshold for mask filtering (where applicable)
florence2_max_tokensoptINT1024256–4096[Florence-2 only] Maximum tokens for generation
florence2_num_beamsoptINT31–5[Florence-2 only] Beam search width
sa2va_max_tokensoptINT2048512–8192[SA2VA only] Maximum tokens for generation
sa2va_num_beamsoptINT11–5[SA2VA only] Beam search width
seedoptINT420–4294967295Fixed seed for reproducible results

Outputs (3)

NameTypeDescription
masksMASK
overlaid_maskIMAGE
textSTRING