Color_detection_histograms

This repository includes:

• a histogram calculator
• a color detection classifier based on histogram numpy arrays (neural networks + svm)
• a color detection classifier based on histogram images (CNN + svm)

Histogram Calculator

your dataset directory should be in this format:

Dataset 
      ├── train
      │   │
      │   ├── blue 
      │   │     ├── image1.jpg
      │   │     └── image2.jpg
      │   │
      │   ├── red 
      │   │     ├── image1.jpg
      │   │     └── image2.jpg 
      │   │
      │   └── yellow
      │         ├── image1.jpg
      │         └── image2.jpg 
      │
      │
      └── test
          │
          ├── blue 
          │     ├── image1.jpg
          │     └── image2.jpg
          │
          ├── red 
          │     ├── image1.jpg
          │     └── image2.jpg 
          │
          └── yellow
                ├── image1.jpg
                └── image2.jpg 

just insert your dataset and output path in create_hist_data.py then run it. it will generate calculated histograms for your dataset.

Color detection

There is 2 way to classifiy your histogram dataset: -as image -as numerical arrays

for each of them there are data_loader.py and classifier.py you can easily insert the path of histograms numpys in dataloader, then run the classifier.