Came up with the universal approach that can increase the accuracy of image classification tasks irrelevant to the domain.
To increase overall precision, I used an ensemble model that combined different image data pre-processing techniques like segmentation, masking, and feature engineering. Convolutional neural networks (CNN) were used in this study to extract features from images and feed them to k-nearest neighbor classifiers (KNNs).
After conducting numerous rigorous experiments, we were able to improve the model's accuracy and precision by 2%, yielding a 99.90% model. Later this approach was also tested on different image classification domains like medical, etc. This approach was presented in the ICACIE21 conference receiving 3 peer-to-peer reviewers.
