A Review and Comprehensive Analysis of Recent Research in Crop Yield Prediction using Machine Learning Algorithms Selvakumar R. K., Salehundam Sai Tanoj, Tigulla Hrushitha Goud Raics IEEE Recent Advances in Intelligent Computational Systems, 2024 Traditional farming based on conventional methods has been the pillar of agricultural practices for a long time. But, due to the escalating global population, unstable climatic patterns and limited resources, the efficiency of traditional methods is facing challenges. The introduction of data-driven and artificial intelligence based approaches involving machine learning and deep learning have gained significant popularity capturing the interest of numerous researchers. This review examines various algorithms, including Random Forest, Decision Trees, K-Nearest Neighbours, Lasso Regression, Stacked Regression, and XGBoost, along with deep learning techniques like LSTM, RNN, and SVM. The study highlights their performance across multiple metrics such as accuracy, RMSE, MAE, R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, and MSE. Notably, Random Forest emerges as a highly effective model due to its excellent performance in accuracy, RMSE, MAE, and R2. Lasso Regression and XGBoost also show remarkable results in RMSE and MSE, respectively. Deep learning models like LSTM, RNN, and SVM are recognized for their high accuracy, underlining their potential in handling complex agricultural data. Through an examination of diverse approaches and data sources, this review seeks to explain and define the capabilities of these innovative technologies for sustainable agrarian applications.
A Survey and Study on 'Vehicle Tracking Algorithms in Video Surveillance System' S. Karuppuchamy, R.K. Selvakumar 2017 IEEE International Conference on Computational Intelligence and Computing Research Iccic 2017, 2018 Vehicle detection is one of the segmentation process after that the detected vehicle in one frame to test whether it is presented in the subsequent frames called vehicle tracking. These concepts are essential for traffic management system. There are number of algorithms are developed by different researchers. But they are considered a specific environment condition for implementing the same. During this study, we identify the problem there is an requirement of an algorithmto detect and track the vehicle in all environmental conditions.
Electronic transaction authentication - A survey on multimodal biometrics S. Shunmugam, R. K. Selvakumar 2014 IEEE International Conference on Computational Intelligence and Computing Research IEEE Iccic 2014, 2014 In this era of Electronic transactions, with the increase in the usage of credit and debit cards, the frauds are also increasing. Various authentication techniques are employed to mitigate the frauds and the biometric authentication plays a vital role. The objective of using biometric authentication is to make sure that the transaction is done by the genuine customer. In this paper, we give a brief discussion on unimodal biometrics, its limitations, need for multimodal biometrics, levels of fusion and multimodal methods adopted in recent works.
A novel approach for image denoising using dynamic tracking with new threshold technique S. Kother Mohideen, S. Arumuga Perumal, N. Krishnan, R. K. Selvakumar 2010 IEEE International Conference on Computational Intelligence and Computing Research Iccic 2010, 2010 During the image acquisition and communication the image is corrupted by noise. This is a classical problem in the field of signal or image processing. In this paper local statistical function is used to find the percentage of noise affected in the noise image. Dynamic denoising algorithm is used simultaneously as a tool for enhancing and tracking image pixel. The pixels are chosen based on the prediction of noise pixels in dynamic tracking. Neighshrink and sure shrink threshold is applied next as a new combination threshold to reduce a noise image in wavelet domain. The PSNR results demonstrated in this paper indicate the many advantages of the proposed dynamic technique using overstate combine Neighshrink and Sureshrink (Neighsureshrink) method.
Compound video image compression for H.264/AVC-INTRA R.K. Selvakumar, Krishnan Nallaperumal, A. Punithavathy Proceedings of the 2008 International Conference on Computing Communication and Networking ICCCN 2008, 2008
