clip-aware-embeddings

What problem does it solve?

This Skill provides accurate semantic image-text matching while intelligently avoiding common pitfalls of general-purpose models like CLIP. It ensures you use the right model for the right task, preventing incorrect results for complex visual queries like counting objects or understanding spatial relationships.

Core Features & Use Cases

• Semantic Image Search: Finds images based on textual descriptions, understanding the meaning behind the words.
• Zero-Shot Classification: Categorizes images into broad categories without needing specific training examples for each category.
• Intelligent Model Selection: Automatically identifies when CLIP is appropriate and when specialized models (e.g., for object counting, fine-grained classification) are needed.
• Use Case: You need to find all images of "a beach at sunset" from a large dataset. This Skill will use CLIP for efficient semantic search. If you then ask "how many people are on the beach?", it will correctly identify that CLIP is not suitable and suggest an object detection model.

Quick Start

Find images that are semantically similar to "a dog playing in a park" from my image library.