Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/Arood/directus-extension-media-ai-bundle
A collection of media-related AI extensions for Directus.
https://github.com/Arood/directus-extension-media-ai-bundle
directus directus-ai-hackathon
Last synced: about 2 months ago
JSON representation
A collection of media-related AI extensions for Directus.
- Host: GitHub
- URL: https://github.com/Arood/directus-extension-media-ai-bundle
- Owner: Arood
- License: mit
- Created: 2023-08-28T15:21:36.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-08-28T16:09:33.000Z (over 1 year ago)
- Last Synced: 2024-04-08T01:17:06.825Z (9 months ago)
- Topics: directus, directus-ai-hackathon
- Language: TypeScript
- Homepage:
- Size: 1.73 MB
- Stars: 58
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-directus - Media AI Bundle - Two operations to perform image description and OCR. (Extensions / Community)
README
# Media AI Bundle
This is a collection of media-related AI extensions for [Directus](https://directus.io), to help you enhance the file library in your next project.
---
## 📋 Details
### ⚡️ Operations
---
#### Describe image
Describe the contents of an image in text form. Useful for creating alt-texts or captions. The format of the returned description varies between different API:s, read below for details.
Required API: [AltText.ai](https://alttext.ai) or [Amazon Rekognition](https://aws.amazon.com/rekognition/)
Successful result
With AltText.ai as API, the `description` will be a more descriptive text:
```json
{
"description": "A cat wearing glasses with red lights on it.", // Image description
"$raw": {...} // The original response from the API
}
```
With Amazon Rekognition as API, the `description` will be a comma-separated list of labels:
```json
{
"description": "Light, Animal, Fish, Sea Life, Shark, Cat, Kitten, Mammal, Pet", // Image description
"$raw": {...} // The original response from the API
}
```
---
#### Extract text from image
Detect text (printed or handwritten) from images and extract them to a single string.
Required API: [Amazon Rekognition](https://aws.amazon.com/rekognition/)
Successful result
> [!NOTE]
> This operation will be triggered as successful even if no text is found. In that case, `full_text` will be an empty string.
```json
{
"lines": [
{
"text": "Lorem ipsum", // Line of text found in an image
"confidence": 99.63353729248047, // How certain the AI is that this match is correct (up to 0.0-100.0)
"geometry": { // Coordinates where the text was found (0.0-1.0)
"top": 0.0693359375,
"left": 0.0615234375,
"height": 0.0869140625,
"width": 0.513671875
}
}
],
"full_text": "Lorem ipsum" // All lines concatenated into a single string,
"$raw": {...} // The original response from the API
}
```
---
## 🛠️ Setup
### Step 1 - Installation
Run: `pnpm install directus-extension-media-ai-bundle`
Or download the release and put it in your `extensions/` folder.
### Step 2 - API keys
Next you need to provide API keys for the services you want to use:
#### AltText.ai
| Variable | Description |
|-------------------------|-------------------------|
| `ALTTEXT_AI_API_KEY` | Your AltText.ai API key |
#### Amazon Rekognition
This extension uses AWS SDK for JavaScript V3 and you might be able to use some of the alternatives listed in [their developer guide](https://docs.aws.amazon.com/sdk-for-javascript/v3/developer-guide/setting-credentials-node.html), but if you run Directus in a Docker environment it might be easiest to configure credentials with environment variables. Refer to [this page](https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credentials_access-keys.html) for information on how to get your access keys.
| Variable | Description |
|-------------------------|-------------------------|
| `AWS_ACCESS_KEY_ID` | Your AWS access key |
| `AWS_SECRET_ACCESS_KEY` | Your secret key |
| `AWS_REGION` | Which [region](https://docs.aws.amazon.com/general/latest/gr/rekognition.html) you want to connect to |
---
## 🔮 Roadmap
- Use Transformations to resize and convert images before sending them to the API.
- Video support where it makes sense.
- Support for other API:s, like Azure Vision AI.
- More operations or other Directus extensions - feel free to send ideas or contribute with your own pull requests.
- More configuration options, such as language, minimum confidence etc.
---
## ❤️ Collaborators
- Arood
- You?