Convolution

People are used to exploring grayscale images in their family albums but it is difficult to grasp the reality without colours. Luckily, with advancements in Machine Learning it has been possible to solve problems previously thought impossible. The authors aim to automatically colourize grayscale images using a subset of Machine Learning called Deep Learning. The system will be trained on an image dataset and given an input grayscale image the model will be able to assign aesthetically believable colours. A grayscale photograph has been provided; our approach solves the problem of visualizing a reasonable colour version of the grayscale picture. This issue is undoubtedly under controlled; therefore earlier methods to this problem have either counted majorly on user interaction or it leads to in unsaturated colourizations. The authors put forward a completely automatic approach that will try to produce realistic and vibrant colourizations as much as possible. The proposed system has been applied as a feed-forward in a Convolutional Neural Network and has been trained on over twenty thousand colour images currently.

Lung infection represents one of the most perilous indicators of Covid-19. The most efficient diagnostic approach entails the analysis of CT scan images. Utilizing deep learning algorithms and machine vision, computer scientists have devised a method for automated detection of this disease. This study proposes a two-stage approach to identifying lung infection. In the initial stage, image features are extracted through a transfer learning framework employing ResNet50, with the last two layers being fixed. Subsequently, a CNN neural network is constructed for image detection and categorization in the second stage. By employing superior image feature selection and minimizing non-informative features, this proposed method achieves impressive accuracy metrics: 98.99% accuracy, 98.91% sensitivity, and 99.10% specificity. Furthermore, a comparative analysis is conducted between this method and six other architectures (Inception, InceptionResNetV2, ResNet101, ResNet152, VGG16, VGG19), with and without transfer learning. The findings demonstrate that the proposed method attains 98% accuracy on test data, without succumbing to overfitting.