Unsupervised Change Detection in Remote Sensing Images Using CNN Based Transfer Learning Josephina Paul 1(B ) , B. Uma Shankar 2 , Balaram Bhattacharyya 1 , and Alak Kumar Datta 1 1 Department of Computer and System Sciences, Visva-Bharati University, Santiniketan 731 235, West Bengal, India 2 Machine Intelligence Unit, Indian Statistical Institute, 203 B.T. Road, Kolkata 700 108, India Abstract. Change detection (CD) using remote sensing images have gained much attention in recent past due to its diverse applications. Devising reliable CD techniques that integrate huge topographical infor- mation is highly challenging. Researches in deep learning paradigm, particularly with Convolutional neural networks (CNN), have proven that CNN are efficient in abstracting knowledge from mul- tiple spectral bands, easy to be trained, and capable of deriving inference from unseen datasets. However, gathering training patterns are difficult in many real life problems and therefore, the pre-trained CNN models can be applied effectively. Hence, we consider three CNN models, VGG19, InceptionV3 and ResNet50 for feature extraction using transfer learning, followed by KMeans and Fuzzy C-Means(FCM) clustering algorithms for generating change maps. The proposed methods have been tested on two repre- sentative datasets of different land cover dynamics and have exhibited promising results with high overall accuracy and Kappa statistic (95.09 & 0.8173 respectively on Dubai city dataset and 97.12 & 0.8970 respec- tively on Texas dataset for Resnet+FCM) as well as superior to the state-of-the-art methods compared. Keywords: Change detection · Convolutional neural networks · Transfer learning · Feature extraction · Clustering 1 Introduction Change detection (CD) using images is aimed at understanding the changes occurred in the same area during an interval of time. Over the past decades the significance of change detection studies has been increasing due to its diverse applications such as medical diagnosis, video surveillance, land use change moni- toring, crop stress detection, urbanization studies and, forest fire and infestations detection [1–5]. A wide array of CD techniques using remote sensing images are c  Springer Nature Switzerland AG 2021 M. Singh et al. (Eds.): ICACDS 2021, CCIS 1440, pp. 463–474, 2021. https://doi.org/10.1007/978-3-030-81462-5_42