Equitable quality of education is still a national strategic issue, especially since the Covid-19 pandemic has not yet ended. Learning through Online Virtual Classroom (OVC) is needed to fulfill equal access to quality education for the community. One of the technologies that can be used to build virtual classrooms online is WebRTC (Web Real-Time Communication). WebRTC is a real-time communication technology or web platform that can be run between browsers without the use of various plug-ins. The purpose of this study is to model the WebRTC topology through the signaling mechanism that works in the construction of the OVC. The identification of the OVC feature is integrated with WebRTC as a real-time communication medium, namely Electronic whiteboard, Screen sharing, File transfer, Recording, Chat room, Calendar Integration, and Moderation. Application is developed using Node.js as backend programming and React.js as frontend programming, which is Socket.io for signaling communication. This study proposes a topology using the signaling mechanism of the npRTC model. The research method is quasi-experimental with a forward engineering approach. System testing is carried out to measure network performance with testing parameters including bandwidth consumption, CPU performance, memory usage, throughput, delay, jitter, and packet loss. The results showed that by using the npRTC signaling mechanism, the CPU load, bandwidth requirements, and large memory usage by the client could be reduced, because throughput was increased, and delay, jitter, and packet loss were reduced. This research on the WebRTC signaling mechanism is that the intermediate server for interactive connectivity establishment using a STUN server does not yet involve a TURN server.

Object classification using image processing simplifies the process. Many approaches have been used to classify the object. In general, classification of mangoes uses image of leaves. In this research, we do a slightly different approach using image of mango itself. Here, two kinds of method are used to classify the object.  Implementations of deep learning using neural network and rule based programming are used in the process. Comparative study of the methods are presented in the article. Our result show that accuracy of deep learning approach is better than the rule based programming. The accuracy is 80% and 8% for neural network and rule based programming, respectively.