Add to Favourites
To login click here
This paper proposes a solution to improve the performance of deep CNN models in real-time applications. The proposed approach uses multi-threshold binarization to extract the vector of discriminative features for classification. The proposed approach shows a significant advantage in accuracy for small datasets, while keeping very close recall score to both deep CNN models for larger datasets. Additionally, the proposed approach provides approximately 5 times lower training and inference time than both ResNet and Ensemble CNN models.
