jwan2 / TemplateRetrieval

Image Retrieval based on VGGNet16 and cosine similarity.

Image Retrieval Engine Based on Keras

Introduction

Image Retrival based on VGGNet16 model application and cosine similarity.

Environment

For Linux or MacOS

1. create conda virtual environment

conda create -n ${your_env_name} python=3.6

2. install libraries in requirements.txt

pip install -r requirements.txt

Usage

Example

Speed and accuracy

• Speed

• Accuracy on big dataset, challenge scenario and challenge card

Challenge type	Speed	Accuracy
100000 templates	ss	first:% second:%
Crop half	ss	first:% second:%
Crop 1/3 with angle	ss	first:% second:%
-90 Orientation	ss	first:% second:%
random Orientation	ss	first:% second:%

• Best dimension choice for retriavel task

Feature Storage

• CSV
• CSV.gzip
• Pickle
• HDF5
• HDF5 zips

Conclusion

1. Pickle is the fastest in read and write, but not usable for big data which causes SystemError.

2. HDF5 is second and easy for data in structure, and also showing a good performance in compressibility.

3. Feather-format could be faster than HDF5 and Pickle.

Update

Todo

Related Paper

Useful linux command

1. file size: du -h --max-depth=1 ${filename}
2. file num: ls -l|grep "^-"| wc -l