Related Experiment Video
Updated: Jun 5, 2026
05:41
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
Published on: February 6, 2020
Sketch-Based Image Retrieval: Benchmark and Bag-of-Features Descriptors
IEEE Transactions on Visualization and Computer Graphics
|December 22, 2010
Summary
We present a new benchmark for evaluating sketch-based image retrieval systems. Our novel descriptors significantly outperform existing methods, with all data publicly available for research.
Area of Science:
- Computer Vision
- Information Retrieval
- Machine Learning
Background:
- Sketch-based image retrieval (SBIR) systems are crucial for searching large image databases using sketches.
- Existing benchmarks may not adequately capture real-world performance for large-scale SBIR.
Purpose of the Study:
- To introduce a comprehensive benchmark for evaluating large-scale sketch-based image retrieval systems.
- To provide a publicly available dataset and methodology for SBIR research.
Main Methods:
- A controlled user study was conducted to collect sketch/image pair matching data.
- New descriptors were developed using the bag-of-features approach.
- The benchmark was used to evaluate the performance of the developed descriptors.
Main Results:
- The developed bag-of-features descriptors significantly outperformed existing descriptors in SBIR.
- The benchmark provides a standardized method for evaluating SBIR system performance.
- The dataset and image database are publicly released.
Conclusions:
- The introduced benchmark is valuable for advancing sketch-based image retrieval research.
- The novel descriptors offer a significant improvement for large-scale SBIR.
- Public data availability will foster further development and comparison in the field.