Face Anonymization

About

This workflow blurs all faces in an image or video.

Capabilities

Face anonymization by blurring detected faces.
Robust face detection in various conditions, including large occlusions and varying lighting conditions.
Suitable for both images and videos.

Known limitations

May miss faces in crowded scenes when too many faces are present.

Recommendations

Best results are achieved with high-resolution images and videos.
On critical applications, implement guardrails to verify anonymization quality.

Model information

Based on a custom fine-tuned version of RF-DETR for face detection.
Implemented a custom task for optimal face blurring.

Usage

Python

First, deploy the workflow.
Install Python library:
```
pip install ikomia-client
```

Integrate the workflow into your Python code:

Process an image

from ikclient.core.client import Client
from ikclient.core.io import ImageIO, StorageObjectIO

with Client(
    url="https://your.scale.endpoint.url",
    token="your-api-token",  # default to os.environ.get("IKOMIA_TOKEN")
) as face_blur:
    results = face_blur.run("path/to/your/image.jpg")
    image_output = results.get_output(assert_type=ImageIO)
    image_output.to_pil().save("blurred_faces_image.jpg")

Process a video

from ikclient.core.client import Client
from ikclient.core.io import StorageObjectIO

with Client(
    url="https://your.scale.endpoint.url",
    token="your-api-token",  # default to os.environ.get("IKOMIA_TOKEN")
) as face_blur:
    results = face_blur.run("path/to/your/video.mp4")
    video_output = results.get_output(assert_type=StorageObjectIO)

    with face_blur.storage.read(video_output["metadata"], streaming=True) as stream, \
         open("blurred_faces_video.mp4", "wb") as out_file:
        for chunk in stream.iter_bytes():
            out_file.write(chunk)

For more details, refer to the documentation.

JavaScript

First, deploy the workflow.
Install JavaScript library:
```
npm install @ikomia/ikclient
```

Integrate the workflow into your JavaScript code:

Process an image

import { Client, ImageIO } from "@ikomia/ikclient";
import fs from "fs/promises";

const client = new Client({
  url: "https://your.scale.endpoint.url",
  token: "your-api-token", // default to process.env.IKOMIA_TOKEN
});

const results = await client.run({
  inputs: [
    "https://your.image.url/image.jpg",
  ],
});
const imageOutput = results.getOutput(0, ImageIO);
await fs.writeFile(
  'blurred_faces_image.png',
  Buffer.from(imageOutput.toArrayBuffer())
);

Process a video

import { Client, StorageObjectIO } from "@ikomia/ikclient";
import fs from "fs/promises";

const client = new Client({
  url: "https://your.scale.endpoint.url",
  token: "your-api-token", // default to process.env.IKOMIA_TOKEN
});

const results = await client.run({
  inputs: [
    "https://your.video.url/video.mp4",
  ],
});

const videoOutput = results.getOutput(0, StorageObjectIO);
const response = await client.storage.read(videoOutput.data.metadata);
await fs.writeFile(
  'blurred_faces_video.mp4',
  Buffer.from(response.arrayBuffer())
);

For more details, refer to the documentation.

REST API

Refer to the documentation for integration using the REST API.