Comparative Evaluation of Hand-Crafted Image Descriptors vs. Off-the-Shelf CNN-Based Features for Colour Texture Classification under Ideal and Realistic Conditions

Abstract

Convolutional Neural Networks (CNN) have brought spectacular improvements in several fields of machine vision including object, scene and face recognition. Nonetheless, the impact of this new paradigm on the classification of fine-grained images—such as colour textures—is still controversial. In this work, we evaluate the effectiveness of traditional, hand-crafted descriptors against off-the-shelf CNN-based features for the classification of different types of colour textures under a range of imaging conditions. The study covers 68 image descriptors (35 hand-crafted and 33 CNN-based) and 46 compilations of 23 colour texture datasets divided into 10 experimental conditions. On average, the results indicate a marked superiority of deep networks, particularly with non-stationary textures and in the presence of multiple changes in the acquisition conditions. By contrast, hand-crafted descriptors were better at discriminating stationary textures under steady imaging conditions and proved more robust than CNN-based features to image rotation.

A comparative study of rice variety classification based on deep learning and hand-crafted features

“…Since this set of rice seed images contain an object and are totally different, it does not outperform CNNbased methods. Similarly, we can see in the conclusion [44] that hand-crafted descriptors outperformed CNN-based features for colour texture classification. In the case of rice seed images, we found the opposite was true based on our experimental results.…”

“…In [42], [43], the authors propose to feed CNN by the new encoded image via Neighbor-Center Difference Image (NCDI) for texture classification. Bello-Cerezo et al [44] present a deep comparative evaluation of hand-crafted and CNN-based features for colour texture classification. The experimental results confirmed that hand-crafted descriptors give better performance than CNN-based features when there was little intra-class variability.…”

A Deep Learning Approach for Mild Depression Recognition Based on Functional Connectivity Using Electroencephalography

“…Convolutional neural networks (CNNs) lie at the core of the best current architecture processing methods for both image and video data, primarily due to the advantages of CNNs in processing 2D input data (Cecotti and Graser, 2011). CNN has good performance in the field of biological image classification, and the features learned from CNN are often better than handcrafted features (Bello-Cerezo et al, 2019). The studies of Nanni et al (2019a;2019b) show that ensemble system of handcrafted and learned features can boost the performance of CNN in bioimage classification.…”

Partial Order Rank Features in Colour Space

“…Before proceeding to definitions it is worth noting that, notwithstanding the trend towards the use of convolutional neural networks as feature extractors [14], rank features are still competitive in texture applications [15]. In the following section we recall the axioms for an order relation.…”