Lung Cancer Classification Using Improvised CNN

State-of-the-art approaches have been enabled by neural networks to attain accurate results on tasks such as detection of objects which are related to computer vision, but the success of these approaches relies on computational resources that are costly and hinders people who prefer cheap devices to advanced technology. A network named Cross Stage Partial Network (CSPNet) is proposed in this paper to diminish the problem that requires computations based on heavy inference in the view of network architecture. This problem is caused due to the duplicate gradient information that is present within the network optimization. The maintenance of the variability of gradients by the proposed networks is done by combining feature maps both at the beginning and the end of a network stage, and the computations is reduced by 20% with equal or even greater accuracy on the image dataset of Chest CT scan. The implementation of CSPNet is quite easy and also standard in nature to deal with architectures that are built on DenseNet and ResNet.