Data Augmentation Method by Applying Color Perturbation of Inverse PSNR and Geometric Transformations for Object Recognition Based on Deep Learning

Abstract

Deep learning is applied in various manufacturing domains. To train a deep learning network, we must collect a sufficient amount of training data. However, it is difficult to collect image datasets required to train the networks to perform object recognition, especially because target items that are to be classified are generally excluded from existing databases, and the manual collection of images poses certain limitations. Therefore, to overcome the data deficiency that is present in many domains including manufacturing, we propose a method of generating new training images via image pre-processing steps, background elimination, target extraction while maintaining the ratio of the object size in the original image, color perturbation considering the predefined similarity between the original and generated images, geometric transformations, and transfer learning. Specifically, to demonstrate color perturbation and geometric transformations, we compare and analyze the experiments of each color space and each geometric transformation. The experimental results show that the proposed method can effectively augment the original data, correctly classify similar items, and improve the image classification accuracy. In addition, it also demonstrates that the effective data augmentation method is crucial when the amount of training data is small.

A deep learning based framework for the registration of three dimensional multi-modal medical images of the head

“…The method used in generating synthetic images from the original dataset and the number of synthetic images to be generated are two important concerns in data augmentation. Many methods have been introduced in response to the former question, such as, random cropping 40 , mixing images 41 , generative adversarial networks 42 , neural style transfer 43 , and geometric transformations 44 , 45 . In our application, our focus is rigid body registration.…”

“…As such, we generated synthetic images by rotating each image by a random angle (in the range of ) around a random axis and translating it by a random distance (in the range of ) along the coordinate axes. Not as much focus has been garnered by how many synthetic images should be generated for optimal training, and typically researchers have used arbitrary numbers that have performed well for their applications 40 , 45 , 46 . We employed an iterative training and testing procedure to identify the amount of synthetic data required to achieve best results.…”

Developing precision agriculture using data augmentation framework for automatic identification of castor insect pests

“…In the case of rotation, each pixel of an image is rotated via its center. It applies the translation of the object between 0° and 360° angles, and the translation of the object changes the values of coordinates ( Kim et al., 2020 ). Flipping generally involves mirroring pixels across the axis (horizontal or vertical).…”

4D Building Reconstruction with Machine Learning and Historical Maps

“…It is most commonly used for dealing with overfitting as well as creating more sample data, e.g., in deep learning processes [54][55][56][57]. In data augmentation, sample sets are expanded, generating synthetic data through any geometric or colorimetric transformation [58,59].…”