Comparing Optimization Algorithms For Enhancing Resnet152v2 Performance

Abstract

Background: While Convolutional Neural Network (CNN) methodologies have widened to include ensemble and the generation of models from the original individual CNN designs, very few research have compared how well these approaches perform when it comes to recognizing and localizing rice diseases. But a lot of people are unaware of the differences between mutton, hog, and beef. The current study uses the ResNet152V2 algorithm to categorize multiple types of beef, mutton, and pork.

Aim: To correctly prepare ResNet152V2, measure the computing assets (e.g., memory use, training time) needed by any optimization procedure.

Method: In the proposed approach, a previously trained Convolutional Neural Networks, architecture is used to extract parameters from a dataset of a mammogram image analysis for left-right comparison. ResNet152V2 is then deployed to distinguish among the four kinds of mammograms (A, B, C, and D).

Results: The results of the trial demonstrate that ResNet152V2 allows for comparison types with an incredible 100% overall accuracy, supporting the model's suitability as a mammography type recognition and classification tool.

Conclusion: Thus, our research provides strong support for the use of ResNet152V2 in the broader context of breast cancer detection and diagnosis.