This is final project of information retrieval
Name | Git | |
---|---|---|
Phạm Xuân Trí | 18521530@gm.uit.edu.vn | Pxtri2156 |
Nguyễn Vương Thịnh | 18520367@gm.uit.edu.vn | ThinhNguyen209 |
Lưu Hoàng Sơn | 18521348@gm.uit.edu.vn | sonlhcsuit |
If you want to query one or more image, you will run this file. And, you can run file to evaluation.
Folder of feature extractions for image : HOG, HSV-Histogram, VGG16, SURF, SIFT, DELF
Folder of similarity measures : Euclidean, Cosine, Manhatan, IOU
To extract feature for dataset. You must run file extract_database.py.
Argument:
- input_folder: The path of folder image that need extract feature.
- output_folder: The path of folder, where are save feature.
- method: method feature extraction that you want to apply.
- LSH: Assign 1, if you want to activate LSH.
python extract_database.py \\
--input_folder <input_path> \\
--output_folder <out_path> \\
--method <feature_method> \\
--LSH 1
Example:
python extract_database.py \\
--input_folder "/content/drive/MyDrive/Information_Retrieval/src/CS336_Final_Project/data/train" \\
--output_folder "/content/drive/MyDrive/Information_Retrieval/src/CS336_Final_Project/feature" \\
--method "HOG" \\
--LSH 1
NOTE: If you are using colab, make sure you add '!' before the command.
To query 1 or more image. You must run file main.py
Argument:
- option: query or eval.
- input_path: the path folder of query image.
- output_path: the path will save result query.
- feature_path: the path of feature dataset.
- feature_method: method feature extraction that you want to apply.
- similarity_measure: method similarity measure that you want to apply.
- ground_truth: the path ground_truth. If you want to query, you can't enter this path.
python main.py \\
--option="query" \\
--input_path=<input_path> \\
--output_path= <out_path> \\
--feature_path= <fearure_path_saved> \\
--feature_method=<extract feature method>\\
--similarity_measure=<compute_similarity_measure> \\
--LSH=1 \\
--ground_truth = None
Example:
python main.py \\
--option="query" \\
--input_path="/content/drive/MyDrive/Information_Retrieval/src/CS336_Final_Project/data/test" \\
--output_path="/content/drive/MyDrive/Information_Retrieval/src/CS336_Final_Project/data/valid" \\
--feature_path="/content/drive/MyDrive/Information_Retrieval/src/CS336_Final_Project/feature" \\
--feature_method="HOG" \\
--similarity_measure="cosine" \\
--LSH=1 \\
--ground_truth = None
If you are not using colab, so change the setup.sh file. Must done, change feature_path in setup.sh into your feature_path.
for colab users, just run (make sure you had mounted drive & data already existed)
bash /content/CS336_Final_Project/setup.sh
python3 app.py -ng 1