Content-based Image Retrieval

Download and Split Data

We use caltech101 dataset to evaluate our system. The dataset can be downloaded from here.

To split train, test and valid set. Please do the following step:

• Extract zip file into data/caltech-101
• Install split-folders python package

pip install split-folders

• Run this python scripts

import splitfolders
splitfolders.ratio('data/caltech-101', output="data/caltech101", seed=1337, ratio=(0.7, 0.15,0.15)) 

Reproduce result

• All results are stored in out/ directory

• Install dependencies: pip install -r requirements.txt

• Run bash.sh to run evaluation on 4 methods of feature extraction: color histogram, sift description, resnet and ensemble.