ICE2607 Lab4: Locality Sensitive Hashing (LSH)

This project uses LSH for retrieving target images from a dataset. It offers a user-friendly command-line interface for image search, allowing comparison of search speeds with the NN algorithm. Additionally, it enables easy plotting of image similarities and utilizes ResNet for enhanced image feature extraction.

File Structure

• src/
  • knn.py: Includes the KNN class.
  • lsh.py: Includes the LSH class.
  • main.py: Main Python script.
  • plot.py: Visualization functions.
  • preprocess.py: Image preprocessing utilities.
• dataset/: Default dataset images.
• results/: Search results and report figures.
• target.jpg: Default target image.
• Lab4_report.pdf
• README.md
• Makefile
• requirements.txt

Install

1. Ensure that you have Python installed on your system.

2. Clone or download this project and navigate to the project directory.

3. Check if the dependencies listed in requirements.txt are installed. If not, create a virtual environment (if necessary) and run make install.

Usage

Default LSH retrieval

python -m src.main \
    --image-dir path/to/image/dataset \
    --target-dir path/to/target/image

Note:

• Avoid duplicate filenames in image-dir.
• If the parameters are not specified, the program will utilize target.jpg as the target image, 40 images in the ./dataset directory.

Retrieval Speed Comparison

python -m src.main \
    --image-dir path/to/image/dataset \
    --target-dir path/to/target/image \
    --type comp

Personalized Projection Set

python -m src.main \
    --image-dir path/to/image/dataset \
    --target-dir path/to/target/image \
    --indicator 1 12 24

Note:

• All numbers in the projection set should fall within [1, 24].

ResNet Feature Extraction

python -m src.main \
    --image-dir path/to/image/dataset \
    --target-dir path/to/target/image \
    --resnet True

Additional: Similarity Plotting

• For common features:

python -m src.plot similarity

• For ResNet features:

python -m src.plot similarity_resnet

Additional: Retrieval Time Exploration

python -m src.plot time_comp

python -m src.plot time_init_search