Skip to content

Latest commit

 

History

History
44 lines (36 loc) · 1.94 KB

README.md

File metadata and controls

44 lines (36 loc) · 1.94 KB
Raw

Build Status

Statistical Pattern Recognition

Labs for University Course

tasks, solutions

Lab 1 - DP algorithm for chain-structured graphical models

Image Denoising (Bernoulli noise)

Examples

python3 lab1/main.py lab1/frequencies.json lab1/alphabet path_to_input_image noise_level

python3 lab1/main.py lab1/frequencies.json lab1/alphabet 'lab1/test_images/hello sweety_0.3.png' 0.3
python3 lab1/main.py lab1/frequencies.json lab1/alphabet 'lab1/test_images/but thence i learn and find the lesson true drugs poison him that so feil sick of you_0.45.png' 0.45

Lab 2 - Min-Sum Diffusion

Image Segmentation

Examples

python3 lab2/main.py input_image alpha n_iter colors

python3 lab2/main.py lab2/test_images/ipt.png 1 10 blue white yellow
python3 lab2/main.py lab2/test_images/map_hsv.png 3 100 blue lime

Lab 3 - Tree Reweighted Message Passing (TRW-S)

Image Inpainting

Examples

python3 lab3/main.py input_image alpha Epsilon n_labels n_iter

python3 lab3/main.py lab3/test_images/mona-lisa-damaged.png 1 0 18 1

Lab 4 - "GrabCut"

NOTE: TRW-S as an energy minimization algorithm (instead of Min-Cut/Max-Flow algorithm)

Interactive Foreground Extraction

Examples

python3 lab4/main.py image_path mask_path gamma n_bg n_fg color_bg color_fg em_n_iter trws_n_iter n_iter 

python3 lab4/main.py lab4/test_images/alpaca.jpg lab4/test_images/alpaca-segmentation.png  50 3 3 blue red 10 10 1
python3 lab4/main.py lab4/test_images/lotus.jpg lab4/test_images/lotus-segmentation.png  50 3 3 lime blue 10 10 1