APP FOR IMAGE RECOGNITION USING NEURAL NETWORKS TECHNIQUES AND DEEP LEARNING.

A Deep Learning approach for image clasification

This work was part of a research which ended up being my final project delivery for my Master's Degree. This project is related to Neurology science applied and Robotics and Automation. Its main purpose was to give support to medical detection of brain tumor since it represents one of the highest rates around Spain of children deaths and actually one of the biggest rates in Europe.

Neural Techniques for image processing.

To deploy this work a VGG16 convolution neural net work architecture was used in which instead of having a large number of hyper-parameter a focused convolution layers of 3x3 filter were prioritized with a stride 1, same padding and maxpool layer of 2x2 filter of stride 2.

This method supports the image clasiffication process through the object detection and classification which is able to classify 1000 images of 1000 different categories with 92.7% accuracy. It is one of the popular algorithms for image classification and is easy to use with transfer learning.

The Outcome

The results after applying VGG16 convolutional neural network showed results highly reliable for 4 different classes of brain tumor (Meningioma, Glioma, Pituitary and No Tumor) which were trained under a public dataset available for researching purposes.

Web APP for users

As the main purpose claims, a web app was developed in order to give an extra value to this researching in which the outcome could be displayed in a web browser, this fact was developed taking in count all the feedback received from the medical expertise and the user experience.