Cross-Modal Retrieval

Coming Soon

This example is planned for a future release. Check back for updates on cross-modal retrieval implementations.

Overview

This example will demonstrate:

• Text-to-image retrieval
• Image-to-text retrieval
• Joint embedding spaces
• Similarity-based ranking

Planned Features

• Dual encoder architectures
• Contrastive learning objectives
• Hard negative mining
• Efficient retrieval with approximate nearest neighbors

Related Documentation

• CLIP Models
• Visual QA
• Image Captioning

References

• Faghri et al., "VSE++: Improving Visual-Semantic Embeddings with Hard Negatives" (2018)
• Lee et al., "Stacked Cross Attention for Image-Text Matching" (2018)