Table of Contents About The Project Built With Getting Started Prerequisites Installation Usage Contributing License Contact
About The Project The project has been done as a part of Phase 2 of CSE 515: Multimedia and Web Database. In this project, you will experiment with • image features, • vector models, • dimensionality curse, • graph analysis
Prerequisites:
- Numpy
- Opencv
- Sklearn
- Skimage
- Json
- Python3.5 or later
- PIL
- PyMongo Library (Mongo Client)
- Mongo DB
Installation:
On ubuntu/linux:
- Pip install -r requirement.txt
Usage: To run:
o Task 1 (Type Classifier) ▪ Given image folder as an input and their corresponding labels image type (for example: cc,con,jitter): ▪ Run Command python Task_1.py ▪ Associates Type Id to the test images
o Task 2 (Subject Classifier)
▪ Given image folder as an input and their corresponding labels subject id (for example: subject id 1-40):
▪ Run Command python Task_2.py
▪ Associates Subject Id to the test images
o Task 3 (Sample Classifier)
▪ Given image folder as an input and their corresponding labels image type, subject id,image sample id (for example: image-X-Y-Z):
▪ Run Command python Task_3.py
▪ Associates Sample Id to the test images
o Task 4 (Locality-Sensitive Hashing)
▪ Give input query image
▪ Number of Hash Functions
▪ Number of Hash Families
▪ Run Command python Task_4.py
▪ Outputs N similar images using LSH algorithm
o Task 5 (VA-Files)
▪ Give input query image
▪ Number of bits
▪ Number of similar images accordingly
▪ Run Command python Task_5.py
▪ Outputs N similar images using VA files algorithm
o Task 6 (Decision-tree-based relevance feedback)
▪ Give input query image
▪ Give feature model
▪ Give Hash function
▪ Hash family
▪ Run Command python Task_6.py
▪ Outputs N similar images and re-ranks the results as per the user's feedback using Decision Tree system.
o Task 7 (SVM-classifier-based relevance feedback)
▪ Give input query image
▪ Give number of bit
▪ Give number of similar image
▪ Give number of feature model
▪ Run Command python Task_7.py
▪ Outputs N similar images and re-ranks the results as per the user's feedback using SVM system.
o Task 8 (Query and feedback interface)
▪ Run Command python Task_8.py
▪ This is used as an interface to execute Task 4,5,6,7