Healthcare Technology

B N M Institute of Technology

Neural Network Based Object Recognition System Using Stereo Images


The conventional methods that perform object recognition use image processing to detect edges and for pruning the image to remove the unwanted background. When object recognition is performed using image processing technique, the images in the database needs pre-processing. However, these are not dynamic and there is no learning or training involved.
Convolutional Neural Networks are known for their image processing capabilities, in that no pre-processing is required before giving an image as input to them. Use of 2D images in object recognition may sometimes result in wrong classification. Since 3D images possess more information, use of such images will result in more reliable classification. In stereoscopic images, the depth information will separate the foreground from the background.
his depth information is used as the 3rd dimension. Neural networks can analyze information the way human brain does. The main object of the invention is to provide a system for real time object recognition using stereoscopic images. It is another object of the invention to provide a system for real time object recognition using convolutional neural network, the said network utilizes three dimensional information obtained from stereoscopic image of the object.

Project Team

Shashanka G, Tejas R Simha, Varun D Gurjar, Vishnuvardhan G
Electronics and Communication Engineering


Dr. Chaitra N
B N M Institute of Technology
Scroll to Top