ResNet vs Inception-v3 vs SVM: A Comparative Study of Deep Learning Models for Image Classification of Plant Disease Detection

One of the primary reasons for decreased crop yields is the presence of plant diseases, which also leads to substantial financial losses for farmers and the entire agricultural industry. It is possible to lessen agricultural losses and stop the spread of diseases by detecting and diagnosing plant diseases early. Deep learning models have shown great potential in recent years to detect and diagnose plant diseases using images. This research summarizes and analyses current state-of-the-art plant disease detection systems that use deep learning. Plant diseases considerably influence crop productivity and quality, leading to economic losses for farmers and food shortages for the people. Because of this, there is a demand for deep-learning models to be used in plant disease identification. Visual inspection and laboratory testing are time-consuming, expensive, and potentially inaccurate. This is especially true early in disease development. Deep learning approaches may solve these issues by automating photo analysis for fast, accurate diagnosis. ResNet, Inception v3, and other deep learning models can learn data representations through several non-linear transformations. They can recognize complicated picture patterns and diagnose diseases accurately. Using variants, deep learning algorithms can learn data representations. This study compares ResNet with Inception-v3, SVM for image classification using the CIFAR-10 dataset. The results of our trials demonstrate that ResNet surpasses Inception-v3 by a significant margin. ResNet achieved a classification accuracy of 99.81%, Inception-v3 only reached 87%, and SVM with 99.88%. We also compare the two models' performance when subjected to varying hyperparameters and find that ResNet is less affected by tweaking these parameters than Inception-v3.