replicatemodel/swinir_t4 – Run with an API on Replicate

replicatemodel / swinir_t4

Cheaper model SwinIR: Image Restoration Using Swin Transformer (analogue of the popular model: jingyunliang/swinir)

Cold

Public
850 runs
T4
GitHub
Paper
License

Run with an API

Playground API Examples README Versions

Input

image

*file

Input image

task_type

string

Image restoration task type

Default: "Real-World Image Super-Resolution-Large"

noise

integer

Noise level, activated for Grayscale Image Denoising and Color Image Denoising. Leave it as default or arbitrary if other tasks are selected

Default: 15

jpeg

integer

Scale factor, activated for JPEG Compression Artifact Reduction. Leave it as default or arbitrary if other tasks are selected

Default: 40

out_format

string

Output image format

Default: "png"

out_jpg_quality

integer

(minimum: 40, maximum: 100)

If the output format is jpg, specify the quality (40...100%)

Default: 95

Run this model in Node.js with one line of code:

npx create-replicate --model=replicatemodel/swinir_t4

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run replicatemodel/swinir_t4 using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "replicatemodel/swinir_t4:6f3a0035cfa43e83811421021d3360750250d6f527fa06f306d870721edc4447",
  {
    input: {
      jpeg: 40,
      image: "https://replicate.delivery/pbxt/KBiA0eZ0VxcpajZa3dhqcbB9Kgr7ZzgEeZPCXt9uuogmW72P/chip.png",
      noise: 15,
      task_type: "Real-World Image Super-Resolution-Large",
      out_format: "png",
      out_jpg_quality: 95
    }
  }
);
console.log(output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run replicatemodel/swinir_t4 using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "replicatemodel/swinir_t4:6f3a0035cfa43e83811421021d3360750250d6f527fa06f306d870721edc4447",
    input={
        "jpeg": 40,
        "image": "https://replicate.delivery/pbxt/KBiA0eZ0VxcpajZa3dhqcbB9Kgr7ZzgEeZPCXt9uuogmW72P/chip.png",
        "noise": 15,
        "task_type": "Real-World Image Super-Resolution-Large",
        "out_format": "png",
        "out_jpg_quality": 95
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run replicatemodel/swinir_t4 using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "6f3a0035cfa43e83811421021d3360750250d6f527fa06f306d870721edc4447",
    "input": {
      "jpeg": 40,
      "image": "https://replicate.delivery/pbxt/KBiA0eZ0VxcpajZa3dhqcbB9Kgr7ZzgEeZPCXt9uuogmW72P/chip.png",
      "noise": 15,
      "task_type": "Real-World Image Super-Resolution-Large",
      "out_format": "png",
      "out_jpg_quality": 95
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

You can run this model locally using Cog. First, install Cog:

brew install cog

If you don’t have Homebrew, there are other installation options available.

Run this to download the model and run it in your local environment:

cog predict r8.im/replicatemodel/swinir_t4@sha256:6f3a0035cfa43e83811421021d3360750250d6f527fa06f306d870721edc4447 \
  -i 'jpeg=40' \
  -i 'image="https://replicate.delivery/pbxt/KBiA0eZ0VxcpajZa3dhqcbB9Kgr7ZzgEeZPCXt9uuogmW72P/chip.png"' \
  -i 'noise=15' \
  -i 'task_type="Real-World Image Super-Resolution-Large"' \
  -i 'out_format="png"' \
  -i 'out_jpg_quality=95'

To learn more, take a look at the Cog documentation.

Run this to download the model and run it in your local environment:

docker run -d -p 5000:5000 --gpus=all r8.im/replicatemodel/swinir_t4@sha256:6f3a0035cfa43e83811421021d3360750250d6f527fa06f306d870721edc4447
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "jpeg": 40,
      "image": "https://replicate.delivery/pbxt/KBiA0eZ0VxcpajZa3dhqcbB9Kgr7ZzgEeZPCXt9uuogmW72P/chip.png",
      "noise": 15,
      "task_type": "Real-World Image Super-Resolution-Large",
      "out_format": "png",
      "out_jpg_quality": 95
    }
  }' \
  http://localhost:5000/predictions

To learn more, take a look at the Cog documentation.

Output

Generated in

8.7 seconds

Tweak itReport

This example was created by a different version, replicatemodel/swinir_t4:3ca651c5.

Examples

View more examples

Run time and cost

This model costs approximately $0.047 to run on Replicate, or 21 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia T4 GPU hardware. Predictions typically complete within 4 minutes. The predict time for this model varies significantly based on the inputs.

Readme

This is a cheaper model SwinIR: Image Restoration Using Swin Transformer, use model video card Nvivdia T4 (16GB) . Нere on the site there is a similar model but more expensive and runs on a more powerful video card Nvivdia A100 (40GB) (analogue: jingyunliang/swinir).

The specified operating time of the model is 4 minutes, which is about 3 minutes starting from a cold state and working for about less than 1 minute.

This model allows you to process large files without error “CUDA out of memory”.

The official repository is here: https://github.com/JingyunLiang/SwinIR

Trained models are taken from here: https://github.com/JingyunLiang/SwinIR/releases/tag/v0.0