Kling Image Recognize
[Core Function] Kling Image Recognize is a vision analysis model. [Strengths] It excels at detecting and extracting semantic information (like fashion/clothing details) from images. [Best For] Highly recommended for: visual search, tagging, and clothing detection. [Limitations] Do NOT use this model to generate images; it only returns text data. [Routing] Use this when the user wants to ‘analyze’, ‘detect’, or ‘recognize’ contents of an image.
Authorizations
API Key authentication. Format: Bearer YOUR_API_KEY.
Body
Note: This is a synchronous model that returns recognition results immediately. Returns 4 types of recognition results: object_seg (main subject), head_seg (face with hair), face_seg (face without hair), cloth_seg (clothing). Result images are valid for 30 days only. This model does NOT support: prompt, negative_prompt, n, aspect_ratio, resolution, image_fidelity, image_reference, human_fidelity, or callback_url parameters.
Image to recognize (URL or Base64 encoded, do not include 'data:' prefix for Base64). Supported formats: .jpg, .jpeg, .png. Maximum size: 10MB. Minimum dimensions: 300x300px. Aspect ratio must be between 1:2.5 and 2.5:1
1"https://example.com/portrait.jpg"