Attendance marking is a common activity to keep track of the presence of students daily in all academic institutions at all grades. Traditional approaches for marking attendance were manual. These approaches are accurate without a chance of marking fake attendance but these are time-consuming and laborsome for a large number of students. To overcome the drawbacks of manual systems, automated systems are developed using radio frequency identification-based scanning, fingerprint scanning, Face-recognition, and Iris scanning based biometric systems. Each system has its pros and cons. Besides, all of these systems suffer from the limitation of human intervention to mark the attendance one by one at a time. To overcome the limitations of existing manual and automated attendance systems, in this work, we propose a robust and efficient attendance marking system from a single group image using face detection and recognition algorithms. In this system, a group image is captured from a high-resolution camera mounted at a fixed location to capture the group image for all the students sitting in a classroom. Next, the face images are extracted from the group image using algorithm followed by recognition using a convolutional neural network trained on the face database of students. We tested our system for different types of group images and types of databases. Our experimental results show that the proposed framework outperforms other attendance marking systems in terms of efficiency and ease of use and implementation. The proposed system is an autonomous attendance system that requires less human-machine interaction, making it possible to easily incorporate in a smart classroom. This project describes multiple attendance system using face recognition algorithm. An robust and efficient face detection based attendance system is implemented using deep learning algorithm. Face recognition widely implemented to recognize the face. Keywords Image editing, image gradient, mean value coordinates, Poisson equation, seamless cloning.