Gemini → Image Understanding

Overview

This action enables you to analyze and understand images using the Gemini API. By providing an image file along with a prompt (your question or instruction), you can instruct Gemini to interpret, describe, or extract insights from the image content using a selected Gemini model.

Inputs

Function Stack

1. Create file resource from image
   • Reads the file payload from the provided image input.
2. Gemini API Request
   • Sends a POST request to:

https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent?key={gemini_api_key}

- The request body includes both the raw image data and the prompt, following Gemini API’s content structure.

3. Precondition
   • Verifies that the response status is 200 to continue processing.
4. Response
   • Returns the result from the response: gemini_api.response.result.

Example Usage

Request

{
  "gemini_api_key": "AIzaSyD...",
  "model": "gemini-1.5-flash",
  "prompt": "Describe what is happening in this image.",
  "image": "(attach image file)"
}

Response

{
  "result": "The image shows a group of people hiking on a mountain trail under a clear sky."
}

Notes

• The model parameter lets you select among available Gemini models for vision tasks (such as gemini-1.5-flash or similar).
• The image input must be an actual image file (PNG, JPEG, etc.).
• Ensure your Gemini API key has the necessary permissions and quota for vision/model usage.
• This action handles encoding and passing the image to Gemini as required by the API.

Troubleshooting

• PERMISSION_DENIED or UNAUTHORIZED: Check your Gemini API key and model permissions.
• INVALID_ARGUMENT: Make sure your prompt is a string and your image is a valid file format.
• UNSUPPORTED_MEDIA_TYPE: Only supported image formats (like jpeg, png) can be processed.
• Other errors: Refer to the Gemini API documentation for detailed troubleshooting guidance.

References

• Gemini API: Content Understanding
• Gemini API: Overview \& Vision Models