Synthetic Data Augmentation of Tomato Plant Leaf using Meta Intelligent Generative Adversarial Network: Milgan

Abstract

Agriculture is one of the most famous case studies in deep learning. Most researchers want to detect different diseases at the early stages of cultivation to save the farmer's economy. The deep learning technique needs more data to develop an accurate system. Researchers generated more synthetic data using basic image operations in traditional approaches, but these approaches are more complicated and expensive. In deep learning and computer vision, the system's accuracy is the crucial component for deciding the system's efficiency. The model's precision is based on the image's size and quality. Getting many images from the real-world environment in medicine and agriculture is difficult. The image augmentation technique helps the system generate more images that can replicate the physical circumstances by performing various operations. It also prevents overfitting, especially when the system has fewer images than required. Few researchers experimented using CNN and simple Generative Adversarial Network (GAN), but these approaches create images with more noise. The proposed research aims to develop more data using a Meta approach. The images are processed using kernel filters. Different geometric transformations are passed as input to the enhanced GANs to reduce the noise and create more fake images using latent points, acting as weights in the neural networks. The proposed system uses random sampling techniques, passes a few processed images to the generator component of GAN, and the system uses a discriminator component to classify the synthetic data created by the Meta-Learning Approach.

An Efficient Multilevel Thresholding Scheme for Heart Image Segmentation Using a Hybrid Generalized Adversarial Network

“…The DICE coefficient is the ratio of the overlapped area multiplied by two intersections and the total area covered by the images. The result does illustrate as shown in equation (2), and it is also known as the "F1-score. "…”

“…GANs are typically measured using the inception score, which assesses how varied the generator's results are (as determined by an image classification, often Inception-v3) or Frechet inception distance (FID). GANs have been suggested as a fast and precise way to predict the generation of high-energy jets [2]. A generator and a discriminator are present in GANs.…”

New trends in detection of harmful insects and pests in modern agriculture using artificial neural networks. a review

“…Data augmentation in the context of CNNs is the process of producing additional training examples by applying various changes to existing pictures in the training dataset (Albanese et al, 2021). Geometric changes such as random rotation, horizontal and vertical flips, random cropping, and transformations such as brightness modifications or color jitter are examples of frequent transformations used for data augmentation in CNNs (Padmanabhuni and Gera, 2022). Adding random cropping can assist the model in learning to distinguish things that are not centered in the image (Genaev et al, 2022).…”

“…For the topic of agricultural pests, this can be done in a variety of ways. There are several methods for creating synthetic data for CNNs (Karam et al, 2022), including generative adversarial networks (GANs), deep learning picture synthesis, data augmentation, and data interpolation (Padmanabhuni and Gera, 2022). Conditional GAN was used by (Abbas et al, 2021) to generate synthetic images for tomato pests and to improve the performances.…”

Hand Gesture Recognition using Deep Learning Models

“…This is done while the images are getting ready to be run through clear winds. This combines tasks including managing, scaling, and flipping [22][23][24][25][26]. This prevented the neural association from being over fitted to a particular quiet picture.…”