Skip to content

image-segmentation

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

image-segmentation

README.md

Image Segmentation Examples

This directory contains examples for various image segmentation models, ranging from general-purpose interactive segmentation (SAM) to specialized tasks like panoptic driving perception (YOLOP).

Examples

rfdetr

cargo run -F cuda-full --example image-segmentation -- rfdetr  --device cuda --processor-device cuda

yoloe-prompt-free

cargo run  -F cuda-full --example image-segmentation -- yoloe-prompt-free  --device cuda --processor-device cuda --ver 26 --scale m

FastSAM

cargo run  -F cuda-full --example image-segmentation -- fastsam  --device cuda --processor-device cuda

YOLOP: Driving Perception

cargo run  -F cuda-full --example image-segmentation -- yolop  --device cuda --processor-device cuda --source images/car-view.jpg

sam

cargo run  -F cuda-full --example image-segmentation -- sam  --device cuda --processor-device cuda --source images/truck.jpg

sam2

# Using module-specific device/dtype for encoder and decoder
cargo run -F cuda-full --example image-segmentation -- sam2 --scale t --encoder-dtype fp32 --encoder-device cuda:0 --decoder-dtype fp32 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg

sam3-tracker

# Single point (using module-specific device/dtype)
cargo run -F cuda-full --example image-segmentation -- sam3-tracker --vision-dtype q4f16 --vision-device cuda:0 --decoder-dtype fp16 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg -p "pos:500,375" 

# Two positive points
cargo run -F cuda-full --example image-segmentation -- sam3-tracker --vision-dtype q4f16 --vision-device cuda:0 --decoder-dtype fp16 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg -p "pos:500,375;pos:1125,625"

# Box prompt (xywh: x, y, width, height)
cargo run -F cuda-full --example image-segmentation -- sam3-tracker --vision-dtype q4f16 --vision-device cuda:0 --decoder-dtype fp16 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg -p "pos:425,600,275,275"

# Box + negative point
cargo run -F cuda-full --example image-segmentation -- sam3-tracker --vision-dtype q4f16 --vision-device cuda:0 --decoder-dtype fp16 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg -p "pos:425,600,275,275;neg:575,750"

# Multiple boxes
cargo run -F cuda-full --example image-segmentation -- sam3-tracker --vision-dtype q4f16 --vision-device cuda:0 --decoder-dtype fp16 --decoder-device cuda:0 --processor-device cuda:0 --source images/truck.jpg -p "pos:75,275,1650,575;pos:425,600,275,275;pos:1375,550,275,250;pos:1240,675,160,75"

Prompt Format

Format Description Example
text Text description -p "cat"
pos:x,y Positive point (2 coords) -p "pos:500,375"
neg:x,y Negative point (2 coords) -p "neg:300,400"
pos:x,y,w,h Positive box (4 coords) -p "pos:480,290,110,360"
neg:x,y,w,h Negative box (4 coords) -p "neg:370,280,115,375"
text;geo;... Text + geometry -p "shoe;pos:480,290,110,360"

Parsing Rules

  1. First part without pos:/neg: prefix → text prompt
  2. Parts with pos:/neg: prefix → geometry (point or box)
  3. 2 coords → point, 4 coords → box (xywh)
  4. If only geometry (no text), "visual" is auto-set

Examples:

  • -p "cat" → text="cat", no geometry
  • -p "pos:480,290,110,360" → text="visual" (auto), 1 positive box
  • -p "shoe;pos:480,290,110,360" → text="shoe", 1 positive box
  • -p "pos:500,375;neg:300,400" → text="visual" (auto), 1 pos point + 1 neg point