TEAM10-PGSSP
Project Title: Learning to Detect Natural Image Boundaries Using Local Brigtness,Color and Texture Cues.
Team Members:
- Aarathi Ramesh Muppalla : 20173018
- Mahesh Pathakoti : 20173022
- Duvvuri Venkatesh : 20173025
- Krishna Sss Tuttagunta : 20173026
This work is carried out as part of SMAI course in IIIT-Hyderabad.
The objective of this work is to detect object boundaries from local images. Matlab is used for this purpose.
Dataset
Berkeley segmentation dataset (BSDS500) https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/segbench/
Features
-
Brightness Gradient
-
Colour Gradient
-
Texture Gradient
Code Overview
All third party code is provided in folder named third_party
| Matlabfile | Description |
|---|---|
| Feature_Extraction.m | Extracts features mentioned above and stores in CSV file |
| Data_Visualisation.m | [visualising the dataset and the extracted features for a sigle training image |
| Feature_Extraction_com.m | Extracts features mentioned above, compress the image and stores in CSV file |
| K_means_com.m | K-means clustering using training data |
| Linear_Regression.m | Implementation of linear regression |
| kmeans_supervised.m | supervised classification using kmeans with Linear Regression |
| nb_com.m | Naive Bayes implementation |
| svm_com.m | SVM implementation |
| rtree_com.m | Classification Tree Implementation |
| PR_com.m | Calculation of precision, Recall and F-measure |
References
- http://www.aces.edu/dept/fisheries/education/pond_to_plate/documents/ExplanationoftheLABColorSpace.pdf
- The Berkeley Segmentation Dataset and Benchmark - https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/bsds/
- Area Under the Precision-Recall Curve: Point Estimates and Confidence Intervals -http://pages.cs.wisc.edu/~boyd/aucpr_final.pdf
- http://davidjohnstone.net/pages/lch-lab-colour-gradient-picker