casia-iva-lab/fastsam 🔢✓📝❓🖼️ → 🖼️

▶️ 37.3K runs 📅 Jun 2023 ⚙️ Cog 0.7.2 🔗 GitHub 📄 Paper ⚖️ License

image-segmentation

About

Fast Segment Anything

Example Output

Output

Performance Metrics

5.04s Prediction Time

5.00s Total Time

All Input Parameters

{
  "iou": 0.9,
  "conf": 0.4,
  "retina": true,
  "box_prompt": "[0,0,0,0]",
  "image_size": 640,
  "model_name": "FastSAM-x",
  "input_image": "https://replicate.delivery/pbxt/J45o004GBhb0mvUJgqQlaMM7lmHkdcOF5e1FV2yFFXttNbCH/download.jpeg",
  "point_label": "[0]",
  "point_prompt": "[[0,0]]",
  "withContours": true
}

Input Parameters

iou Type: numberDefault: 0.7: iou threshold for filtering the annotations
conf Type: numberDefault: 0.25: object confidence threshold
retina Type: booleanDefault: true: draw high-resolution segmentation masks
box_prompt Type: stringDefault: [0,0,0,0]: [x,y,w,h]
image_size Default: 640: Choose image size
model_name Default: FastSAM-x: choose a model
input_image (required) Type: string: Input image
point_label Type: stringDefault: [0]: [1,0] 0:background, 1:foreground
text_prompt Type: string: use text prompt eg: "a black dog"
point_prompt Type: stringDefault: [[0,0]]: [[x1,y1],[x2,y2]]
withContours Type: booleanDefault: false: draw the edges of the masks
better_quality Type: booleanDefault: false: better quality using morphologyEx

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

image 1/1 /tmp/tmpbn30lvdudownload.jpeg: 448x640 90 objects, 89.9ms
Speed: 4.1ms preprocess, 89.9ms inference, 50.3ms postprocess per image at shape (1, 3, 640, 640)

Version Details

Version ID: 371aeee1ce0c5efd25bbef7a4527ec9e59188b963ebae1eeb851ddc145685c17
Version Created: June 30, 2023

Run on Replicate →