Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Few Shots Text to Image Retrieval: New Benchmarking Dataset and Optimization Methods

Ofer Idan, Vladi Vexler, Gil Lederman, Dima Sivov, Aviad Cohen Zada, Shir Niego Komforti

Abstract

Pre-trained vision-language models (VLMs) excel in multimodal tasks, commonly encoding images as embedding vectors for storage in databases and retrieval via approximate nearest neighbor search (ANNS). However, these models struggle with compositional queries and out-of-distribution (OOD) image-text pairs. Inspired by human cognition's ability to learn from minimal examples, we address this performance gap through few-shot learning approaches specifically designed for image retrieval. We introduce the Few-Shot Text-to-Image Retrieval (FSIR) task and its accompanying benchmark dataset, FSIR-BD - the first to explicitly target image retrieval by text accompanied by reference examples, focusing on the challenging compositional and OOD queries. The compositional part is divided to urban scenes and nature species, both in specific situations or with distinctive features. FSIR-BD contains 38,353 images and 303 queries, with 82% comprising the test corpus (averaging per query 37 positives, ground truth matches, and significant number of hard negatives) and 18% forming the few-shot reference corpus (FSR) of exemplar positive and hard negative images. Additionally, we propose two novel retrieval optimization methods leveraging single shot or few shot reference examples in the FSR to improve performance. Both methods are compatible with any pre-trained image encoder, making them applicable to existing large-scale environments. Our experiments demonstrate that: (1) FSIR-BD provides a challenging benchmark for image retrieval; and (2) our optimization methods outperform existing baselines as measured by mean Average Precision (mAP). Further research into FSIR optimization methods will help narrow the gap between machine and human-level understanding, particularly for compositional reasoning from limited examples.

Few Shots Text to Image Retrieval: New Benchmarking Dataset and Optimization Methods

Abstract

Pre-trained vision-language models (VLMs) excel in multimodal tasks, commonly encoding images as embedding vectors for storage in databases and retrieval via approximate nearest neighbor search (ANNS). However, these models struggle with compositional queries and out-of-distribution (OOD) image-text pairs. Inspired by human cognition's ability to learn from minimal examples, we address this performance gap through few-shot learning approaches specifically designed for image retrieval. We introduce the Few-Shot Text-to-Image Retrieval (FSIR) task and its accompanying benchmark dataset, FSIR-BD - the first to explicitly target image retrieval by text accompanied by reference examples, focusing on the challenging compositional and OOD queries. The compositional part is divided to urban scenes and nature species, both in specific situations or with distinctive features. FSIR-BD contains 38,353 images and 303 queries, with 82% comprising the test corpus (averaging per query 37 positives, ground truth matches, and significant number of hard negatives) and 18% forming the few-shot reference corpus (FSR) of exemplar positive and hard negative images. Additionally, we propose two novel retrieval optimization methods leveraging single shot or few shot reference examples in the FSR to improve performance. Both methods are compatible with any pre-trained image encoder, making them applicable to existing large-scale environments. Our experiments demonstrate that: (1) FSIR-BD provides a challenging benchmark for image retrieval; and (2) our optimization methods outperform existing baselines as measured by mean Average Precision (mAP). Further research into FSIR optimization methods will help narrow the gap between machine and human-level understanding, particularly for compositional reasoning from limited examples.

Paper Structure

This paper contains 30 sections, 6 equations, 9 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Vision Language Foundation Models
  4. Weaknesses of VLMs
  5. Composed Image Retrieval
  6. Robust fine-tuning
  7. Few-Shot Prompt Learning
  8. FSIR Benchmarking Dataset (FSIR-BD)
  9. FSIR-BD Sub-datasets
  10. FSIR-BD-Compositional-VG - Urban-life
  11. FSIR-BD-Compositional-INQUIRE
  12. FSIR-BD-OOD
  13. Analysis and Quality Assessment
  14. FSIR Retrieval Optimization Methods
  15. K-Shot One-Way Prompt Learning (FSIR-PL)
  16. ...and 15 more sections

Figures (9)

  • Figure 1: Example of FSIR-BD. Per query text we have multiple positive and hard-negative images.
  • Figure 2: BLIP scores fall within [0.1,0.5]; learnable scaling factors a,b map these to the full BCE range [0,1].
  • Figure 3: FSIR-CTR training architecture. The reference image plus text query are input to the MLLM, which fuse them and align the result to the external image encoder's embedding space.
  • Figure 4: Query text plus reference image vs. target images from FSIR-TD.
  • Figure 5: Example pf FSIR-BD-Compositional-VG data. Per text query we provide multiple positive images and negative images.
  • ...and 4 more figures