IJSRSET2073101 | Accepted : 05 June 2020 | Published : 16 June 2020 | May-June 2020 [ 7 (3) : 394-400 ] International Journal of Scientific Research in Science, Engineering and Technology (www.ijsrset.com) © 2020 IJSRSET | Volume 7 | Issue 3 | Print ISSN: 2395-1990 | Online ISSN : 2394-4099 DOI : https://doi.org/10.32628/IJSRSET 394 Melanoma Cancer Detection using Deep Learning Megha Gaikwad, Pooja Gaikwad, Priyanka Jagtap, Saurabh Kadam, Prof. Rashmi R. Patil Department of Computer Engineering, Navsahyadri Education Society’s Group of Institutions, Pune, Maharashtra, India ABSTRACT Now a days, skin cancer is well known reason for human death. abnormal skin cells growth is known as skin cancer ,these skin cells generated on human body which exposed to the sunlight, it can generate anywhere on the human body. At early stage, most of the cancers are curable. Hence, it is required to detect skin cancer at early stage to save patient life. It is possible to recognise skin cancer at early stage with advanced technology. Here we present a novel framework using deep learning method and a local descriptor encoding strategy for recognition of dermoscopy image. In particular, the deep representations of a rescaled dermoscopy image first extricated through an exceptionally deep residual neural network, which is pre-trained on a large natural image dataset. After that, local deep descriptors are collected by order less visual statistic features depends on fisher vector encoding to build a global image representation. At last utilized the fisher vector encoded representations to arrange melanoma images utilizing a convolution neural network (CNN). This proposed system is able to generate more discriminative features to deal with large variations within melanoma classes as well as small variations among melanoma and non-melanoma classes with limited training data. Keywords : Dermoscopic Image Recognition, Cnn Algorithm, Melanoma Detection, Segementation. I. INTRODUCTION It is difficult task even for experienced dermatologists to predict skin lesions because of a little difference between encompassing skin and injuries, the visual likeness between skin sores, stupefied lesion outskirt, and so forth. To diagnose cancerous skin lesions at the earliest stage an automated computer-aided detection system with given images can help clinicians. The advancement in deep learning consist of dilated convolution known to have enhanced accuracy with the similar amount of computational complexities as compared with traditional CNN. To give proper treatment recognition of skin lesion is important. Hence, the survival rate is increased due to early recognition of melanoma in dermoscopic images. The accurate detection of melanoma skin lesions is possible to highly trained dermatologists. Therefore, it is very challenging task to detect melanoma due to little difference among lesions and skin, visual similarity between melanoma and non-melanoma lesions, etc. As expertise is in limited supply, reliable automatic detection of skin tumours i.e. a system that can automatically analyse skin lesions, will be very advantageous to enhance the accuracy and efficiency of pathologists. Overall, to tackle these issues, here presented a framework to locate the challenges for automated and accurate melanoma detection in dermoscopy images. The contributions of this study is two-folded. Based on the deep CNN and feature encoding strategy proposed an efficient framework. Is helpful to