twn39/lama 🖼️ → 🖼️

▶️ 39.5K runs 📅 Oct 2023 ⚙️ Cog 0.8.6 📄 Paper
image-inpainting image-object-removal

About

🦙 LaMa Image Inpainting, Resolution-robust Large Mask Inpainting with Fourier Convolutions, WACV 2022

Example Output

Output

Example output

Performance Metrics

7.06s Prediction Time
7.10s Total Time
All Input Parameters
{
  "mask": "https://replicate.delivery/pbxt/JvTNNI5GXp1iol2Sjf3XYfUX83Bemhx4Q5TwoAOeZJXrvc8m/image_inpainting_mask.png",
  "image": "https://replicate.delivery/pbxt/JvTNNPkuMNTMjetsF4LO9eKdXAPy9xkNqUcphkbFcA2WF2TH/image_inpainting.png"
}
Input Parameters
mask (required) Type: string
Mask input image
image (required) Type: string
Origin input image
Output Schema

Output

Type: stringFormat: uri

Example Execution Logs
Original image too large for refinement! Resizing (999, 1499) to (774, 1162)...
  0%|          | 0/15 [00:00<?, ?it/s]
  0%|          | 0/15 [00:00<?, ?it/s]
Refining scale 2 using scale 1 ...current loss: 0.1006:   0%|          | 0/15 [00:00<?, ?it/s]
Refining scale 2 using scale 1 ...current loss: 0.1006:   7%|▋         | 1/15 [00:00<00:03,  3.54it/s]
Refining scale 2 using scale 1 ...current loss: 0.0888:   7%|▋         | 1/15 [00:00<00:03,  3.54it/s]
Refining scale 2 using scale 1 ...current loss: 0.0888:  13%|█▎        | 2/15 [00:00<00:03,  3.30it/s]
Refining scale 2 using scale 1 ...current loss: 0.0810:  13%|█▎        | 2/15 [00:00<00:03,  3.30it/s]
Refining scale 2 using scale 1 ...current loss: 0.0810:  20%|██        | 3/15 [00:00<00:03,  3.23it/s]
Refining scale 2 using scale 1 ...current loss: 0.0761:  20%|██        | 3/15 [00:01<00:03,  3.23it/s]
Refining scale 2 using scale 1 ...current loss: 0.0761:  27%|██▋       | 4/15 [00:01<00:03,  3.20it/s]
Refining scale 2 using scale 1 ...current loss: 0.0722:  27%|██▋       | 4/15 [00:01<00:03,  3.20it/s]
Refining scale 2 using scale 1 ...current loss: 0.0722:  33%|███▎      | 5/15 [00:01<00:03,  3.18it/s]
Refining scale 2 using scale 1 ...current loss: 0.0692:  33%|███▎      | 5/15 [00:01<00:03,  3.18it/s]
Refining scale 2 using scale 1 ...current loss: 0.0692:  40%|████      | 6/15 [00:01<00:02,  3.17it/s]
Refining scale 2 using scale 1 ...current loss: 0.0663:  40%|████      | 6/15 [00:02<00:02,  3.17it/s]
Refining scale 2 using scale 1 ...current loss: 0.0663:  47%|████▋     | 7/15 [00:02<00:02,  3.16it/s]
Refining scale 2 using scale 1 ...current loss: 0.0640:  47%|████▋     | 7/15 [00:02<00:02,  3.16it/s]
Refining scale 2 using scale 1 ...current loss: 0.0640:  53%|█████▎    | 8/15 [00:02<00:02,  3.16it/s]
Refining scale 2 using scale 1 ...current loss: 0.0618:  53%|█████▎    | 8/15 [00:02<00:02,  3.16it/s]
Refining scale 2 using scale 1 ...current loss: 0.0618:  60%|██████    | 9/15 [00:02<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0600:  60%|██████    | 9/15 [00:02<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0600:  67%|██████▋   | 10/15 [00:03<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0582:  67%|██████▋   | 10/15 [00:03<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0582:  73%|███████▎  | 11/15 [00:03<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0567:  73%|███████▎  | 11/15 [00:03<00:01,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0567:  80%|████████  | 12/15 [00:03<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0553:  80%|████████  | 12/15 [00:03<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0553:  87%|████████▋ | 13/15 [00:04<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0540:  87%|████████▋ | 13/15 [00:04<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0540:  93%|█████████▎| 14/15 [00:04<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0528:  93%|█████████▎| 14/15 [00:04<00:00,  3.15it/s]
Refining scale 2 using scale 1 ...current loss: 0.0528: 100%|██████████| 15/15 [00:04<00:00,  3.69it/s]
Version Details
Version ID
2b91ca2340801c2a5be745612356fac36a17f698354a07f48a62d564d3b3a7a0
Version Created
December 8, 2023
Run on Replicate →