ABSTRACT
This report is on topic of simulation and analysis of different heating method for bio-digester substrate. Now-a-days the energy demand is increasing so we have to look other options and devise a method to optimize the production from other sources. Due to Covid-19 mass migration and increased hospital admission occurs, to fulfill the food supply biogas is sought. This report focus on digesters on a small scale that can be employed for household activities. To increase the biogas yield among different influencing factors temperature is chosen and worked upon. Along with insulation there is a heating method installed to maintain the stable temperature which facilitates breakdown of organic materials and improve the productivity. In colder climates maintaining mesophilic temperature can be a challenge, therefore three heating methods are simulated and analyzed. The study reveals about floor heating, in-vessel heating and floor + in-vessel heating method. In-vessel heating method provides uniform cooling, whereas floor heating can be applied at relatively cold climates because it give significant temperature rise (about ∼14 °C). Out of these three methods floor + in-vessel heating method is found suitable as it optimize the benefits of both floor and in-vessel heating methods with 5 °C temperature raise. © 2022
ABSTRACT
Coronavirus (Covid-19) is continuously spreading all over the world now with different variances like alpha, beta, gamma, Epsilon, and Zeta. There are numerous careful steps are taken to minimize the spread of corona virus. Wearing face mask is one among the important measures that people should follow but unfortunately the majority of individuals are not following this safety measure in public spaces. In this paper, we have proposed a novel face mask detection model by cascading three CNN models such as YOLO V3, Facenet and Mobilenet. Here, YOLO V3 model is used for detecting people in a surveillance video data which helps to reduce the search space for face detection techniques. Second CNN model namely Facenet is incorporated to detect the face which is then fed to Mobilenet for mask detection. In the proposed work, mask detection is considered as a binary class problem where a people without mask are discriminated from the people with face mask. A transfer learning is adopted for YOLO V3 and Facenet models. The output layer of Mobilenet is modified for the binary classification. Mobilenet is trained with 3833 instances belonging to both the classes were collected from a realtime data acquired in the laboratory environment during offline classes and from sources like Kaggle and Google images. The trained model has achieved 99.2% accuracy for unseen data. © 2021 IEEE.