DeepLumina: A Method Based on Deep Features and Luminance Information for Color Texture Classification

Table 1

DeepLumina framework.

---

DeepLumina -proposed method

---

Step 1. Resize the input RGB images and perform color space transformation to the YIQ model

Step 2. Estimation of luminance images (Y channel images) from YIQ color model

Step 3. Apply step 1 and step 2 images to the pretrained ResNet101 model

Step 4. Extraction of deep or learned feature map from the ResNet101

Step 5. Classification using a fast linear solver support vector machine

---