https://github.com/motemen/macos-obs-websocket-ocr

A proxy for obs-websocket that adds Optical Character Recognition (OCR) capabilities.
https://github.com/motemen/macos-obs-websocket-ocr

macos obs vision-framework

Last synced: about 2 months ago
JSON representation

A proxy for obs-websocket that adds Optical Character Recognition (OCR) capabilities.

• Host: GitHub
• URL: https://github.com/motemen/macos-obs-websocket-ocr
• Owner: motemen
• Created: 2024-06-18T16:05:42.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-06-28T04:01:18.000Z (7 months ago)
• Last Synced: 2024-11-26T18:28:31.439Z (about 2 months ago)
• Topics: macos, obs, vision-framework
• Language: Swift
• Homepage:
• Size: 60.5 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# OBS WebSocket OCR Proxy for macOS

A proxy for obs-websocket that adds Optical Character Recognition (OCR) capabilities.

Utilizes macOS’s [Vision framework](https://developer.apple.com/documentation/vision) to perform OCR on captured screenshots.

Currently, it introduces one special request type: `__GetTextFromLastScreenshot`. See below for details.

## Usage

    obs-websocket-ocr [--upstream-url URL] [--port PORT] [--hostname HOSTNAME]

- `--upstream-url URL`: The URL of the upstream obs-websocket server. Default: `ws://localhost:4455`.

- `--port PORT`: The port to bind to. Default: `4456`.

- `--hostname HOSTNAME`: The hostname to bind to. Default: `localhost`.

When started, the proxy will listen on `localhost:4456` and forward all messages to the upstream obs-websocket server.

You can connect to the proxy using the obs-websocket client as you would with a normal obs-websocket server.

## Request types

In addition to the [standard obs-websocket request types](https://github.com/obsproject/obs-websocket/blob/master/docs/generated/protocol.md#getversion), the proxy adds one special request type: `__GetTextFromLastScreenshot`.

### \_\_GetTextFromLastScreenshot

Does OCR on the last screenshot taken by OBS and returns the recognized text items and the bounding boxes of them.

#### Response fields:

| Name           | Type                | Description                |

| -------------- | ------------------- | -------------------------- |

| `text_results` | `Array` | The recognized text items. |

TextResult:

| Name           | Type                 | Description                   |

| -------------- | -------------------- | ----------------------------- |

| `text`         | `string`             | The recognized text.          |

| `bounding_box` | `Array` | The bounding box of the text. |

BoundingBox:

| Name     | Type     | Description                              |

| -------- | -------- | ---------------------------------------- |

| `x`      | `number` | The x-coordinate of the top-left corner. |

| `y`      | `number` | The y-coordinate of the top-left corner. |

| `width`  | `number` | The width of the bounding box.           |

| `height` | `number` | The height of the bounding box.          |

## Author

Hironao Otsubo (motemen)