Classification and Measuring Accuracy of Lenses Using Inception Model V3

Convolution network is widely used nowadays for the most computer vision solutions. Deep convolution network has become mainstream and is yielding various benchmarks since 2014. Deep learning has shown tremendous results in the field of image processing. Artificial neural networks are inspired by biological neurons and allow us to map various complex functions at a very fast speed. These neural networks have to train on large sets of data and then take a certain amount of time to do so. Eventually, they give immediate quality gains for most of the tasks in use cases of lens vision and big data scenarios. In this paper, Inception model V3 has been used that aims at utilizing the added computation as efficiently as possible in case of lens images. This model takes bulk images of lenses and fragments them into six different classes depending upon their properties. Inception model achieves accuracy for recognizing images with more than 1000 classes. Histogram shows the comparative study of six different classes on basis of classification rate and accuracy. Confusion matrix has been made which gives the correct predictions for the images of lens. Finally, the results have been compiled in two parts. In the first part contain, we find out the precision and recall of lens images. In the second part, overall accuracy of the model has been compiled which comes out to be 95.9%.