Dissection of Deep Learning with Applications in Image Recognition

Deep Learning has dramatically improved the accuracy of image recognition. Image recognition is considered to be one of the most challenging problems in imaging science. This paper briefly surveys the application of Deep Learning in classification, object detection, semantic segmentation and instance segmentation of objects in digitized images. The paper reviews widely used Deep Learning models, in classifying, detecting and segmenting objects based on convolutional neural networks (CNN). It provides a summary and presents some discussion about the performance of each model, including training time and accuracy.