The article discusses the development of an AI-based system for automatic no-ball detection in cricket with an accuracy of 0.98. The system uses machine learning and deep learning techniques, including Convolutional Neural Networks (CNN) and transfer learning with models like VGG16 and VGG19. The study involved creating a dataset, cropping images to focus on the bowler's end, applying image enhancement techniques, and training the models. The proposed framework aims to assist on-field umpires and reduce human error in critical decision-making during cricket matches. The study also highlights the potential impact of such technology on fan experience and engagement. Despite the success of the system, future work could involve real-time detection from video frames, training on larger datasets, and integrating with cricket ground cameras.

The article discusses a study on classifying osteosarcoma images using machine learning (ML) and deep learning (DL) techniques. The study was conducted in Google Colaboratory with NVIDIA Tesla K80 graphics card and utilized Python, PyTorch, Keras, and other libraries. The study aimed to classify images into non-tumor, necrosis, and viable tumor phases using various classifiers and feature extractors like DenseNet-121, VGG-16, VGG-19, ResNet-50, and Xception. The MLP classifier combined with DenseNet-121 achieved the highest accuracy. The study also explored optimization algorithms, with Adam outperforming SGD and Lbfgs in terms of convergence and efficiency. Feature selection techniques were also analyzed, with Recursive Feature Elimination (RFE) providing the best results. The study concluded that DenseNet-121 as a feature extractor and MLP as a classifier, with RFE for feature selection, was the most effective combination for classifying osteosarcoma images.