Sentimagi Python Image Analysis Library

Requirements

sudo apt-get install python-skimage sudo pip install svgwrite sudo apt-get install python-pywt

General

This library can be used for general image classification and feature extraction.

Feature extraction:

Extract and plot features from a single file

python featureExtraction.py -featuresFile sampledata/spectrograms/music/m_5_r_139.png

Extract features from two files and compare

python featureExtraction.py -featuresFilesCompare sampledata/spectrograms/music/m_5_r_139.png sampledata/spectrograms/speech/kill_bill_2_speech_17.png

Extract features from a set of images stored in a folder

python featureExtraction.py -featuresDir sampledata/spectrograms2/music/

Extract features from a set of directories, each one defining an image class

python featureExtraction.py -featuresDirs spectrograms sampledata/spectrograms/music sampledata/spectrograms/speech

(Features are stored in file "sectrograms_features")

Training and testing classification - regression models:

Train an image classification model

Models are trained from samples stored in folders (one folder per class).

Examples:

• kNN model training

python train.py -train knn knnSpeechMusicSpecs  sampledata/spectrograms/music sampledata/spectrograms/speech

The above example trains a kNN classification model, does cross validation to estimate the best parameter (k value) and stores the model in a file (named knn3Classes).

• SVM model training

python train.py -train svm svmSpeechMusicSpecs  sampledata/spectrograms/music sampledata/spectrograms/speech

The above example trains an SVM classification model, does cross validation to estimate the best parameter (C value) and stores the model in a file (named svmSentimentAds).

Classify an unknown image examples

python train.py -classifyFile knn knnSpeechMusicSpecs sampledata/music.melodies_snatch_0081.png
python train.py -classifyFile knn knnSpeechMusicSpecs sampledata/s_30_r_335.png