Hand-Drawn Electronic Component Recognition Using ORB
S Pavithra, N K Shreyashwini, H S Bhavana, Ganivada Nikhitha, T. Kavitha
Abstract
There is growth in interest to build a system that can recognize freely drawn symbols. In past two years, the situation in the world has become more difficult, almost all the work in their organizations has been digitalized. However, many challenges remain in terms of the recognition accuracy of different drawing styles. To address these challenges, a new approach is proposed to classify and predict the hand-drawn component. In this work Computer Vision approach using the ORB algorithm is used to recognize and predict the component. The drawing tools that are being used need to be upgraded in order to have a better learning experience in order to recognize the symbol that is drawn and directly convert it into digitized form. As it is very difficult to write it on an online platform every time this method would be very useful. Here 15 different electronic components are considered with three different orientations. In this approach, GUI is used to draw the symbol which is easy instead of picking and placing. This input-drawn image is given to the model developed using the CV algorithm, where the input image gets compared with the database image, according to ORB using the FAST method key points are generated for both images. Using a Brute force matcher both images get matched depending on the number of matched features, and the output image gets predicted. The results of the ORB and SIFT CV algorithms are analyzed and compared.