Using 3 different images form to chose the best way to recognize moviments.
Model | Mean Accuracy (%) | AUC | Training Time (min) |
---|---|---|---|
CNN Model (created) | 99,829 | 0,998 | 9,806 |
VGG 16 | 97,027 | 0,989 | 15,070 |
VGG 19 | 94,917 | 0,969 | 32,372 |
RESNET 50 | 96,922 | 0,979 | 28,835 |
INCEPTION V3 | 97,783 | 0,973 | 29,552 |
Created by Wyctor Fogos da Rocha- 29/05/2021.
Some references:
- M. Ribeiro and J. Samatelo. "Traffic Flow Classification using a Video Descriptor and a Convolutional Neural Network", in Anais do XV Workshop de Visão Computacional, São Bernardo do Campo, 2019, pp. 137-142, doi: https://doi.org/10.5753/wvc.2019.7642.
- Dynamic Gesture Recognition by Using CNNs and star RGB: a Temporal Information Condensation. doi:https://doi.org/10.1016/j.neucom.2020.03.038