Publications

Abstract : In this paper, an effective deep learning validation model to improve the quality of dataset under study and to enhance model performance in order to predict deficiencies in tomato plants was developed (Ganesan et al. (2021) Fault detection in satellite power system using convolutional neural network). This solution framework eliminates the background influence in the image plane and differences in lighting conditions and stays true to picturize whatever characteristics are synonymous to each class the model has learned from its training stage. It is possible to obtain positive financial and environmental benefits from these activities by carefully utilizing the above modern computer vision techniques (Sakthivel et al. (2010) Application of support vector machine (SVM) and proximal support vector machine (PSVM) for fault classification of monoblock centrifugal pump. Int J Data Anal Tech Strat). Mathematical numbers can be misleading at times like the model can be prone to overfitting failing to generalize well to unknown test images. So, the best way to validate our model is to form a visualization mechanism for the user (Hari et al. (2022) Fault detection in SPS using image encoding and deep learning [3]). Agriculture is a field that is largely deprived of visualizing deep networks (https://www.kdnuggets.com/2020/06/crop-disease-detection-computer-vision.htm,) as the complexity that comes with it is huge. The techniques of DeepDream and Gradient Class Activation Mapping proposed here help us to solve the issue, which directly meets the flaw. The advantages that we can extract from these validation techniques are that they can assist us in comprehending whether our model has learned the fundamental orientations in our dataset properly, whether the training procedure should be re-evaluated (Soni et al. (2021) Deep learning-based approach to generate realistic data for ADAS applications. In: 2021 5th international conference on computer, communication and signal processing (ICCCSP), and also help us understand whether we may be in need to collect furthermore data for improved performance.