Skip to main content

Keypoint tracking and classification

This project is devoted to tracking and classification of the anchor points during the performance of Transcatheter Aortic Valve Replacement.
Created on April 10|Last edited on March 30

Section 1: Project description



Section 2: Neural networks description



Section 3.1: MobileNet V2 (training)



Section 3.2: MobileNet V2 (convergence)



Section 3.3: MobileNet V2 (prediction)



Section 4.1: ResNet V2 (training)



Section 4.2: ResNet V2 (convergence)



Section 4.3: ResNet V2 (prediction)



Section 5.1: Inception V3 (training)



Section 5.2: Inception V3 (convergence)



Section 5.3: Inception V3 (prediction)



Section 6.1: Inception ResNet V2 (training)



Section 6.2: Inception ResNet V2 (convergence)



Section 6.3: Inception ResNet V2 (prediction)



Section 7.1: EfficientNet B5 (training)



Section 7.2: EfficientNet B5 (convergence)



Section 7.3: EfficientNet B5 (prediction)



Section 8.1: Model comparison (accuracy)



Section 8.2: Model comparison (performance)



Section 9: Comparison of best 3 models