In recent times, since smart devices are growing rapidly all around the globe, so are the Android-based applications that are available for free. This has resulted in a surge of malicious activities. The disastrous effects of this are seen on security and user privacy. Identifying Malware on Android devices is a concern that is on the rise due to the unwanted resemblance between normal and malicious features. This affects the timely detection and malware remains undetected on the devices for a longer time. The popular target of the attackers has become Android devices due to the open nature and popularity of the platform. Since the Android market is growing, it is important to design tools that prove to be efficient in identifying malware attacks on Android-based platforms. Hence, this research work discloses a novel framework for identifying malware in the Android implemented applications with the help of Deep Learning methodologies such as Convolutional Neural Network (CNN) based DenseNet 169 and 201 with Transfer Learning. It has 169 and 201 layers respectively and is composed of several building blocks, including dense blocks, transition layers, and convolutional layers. Plus, using Transfer Learning, we have reduced the time and the computational power the model would have otherwise taken to identify malware in the input file, leading to improved training efficiency and accuracy. The proposed approach was tested and trained using Malevis and Malimg Dataset. The results obtained from the experimental procedures demonstrate that the proposed deep learning framework outperforms the various traditional methods with an accuracy of 98.40% and 96.09% respectively, thus having better accuracy at detecting malware in Android-based applications.