From: Skin lesion classification system using a K-nearest neighbor algorithm
Applied method or technique | Accuracy of measurements (%) | Remarks | Reference |
---|---|---|---|
CNN | 90.0 | Skin diseases can be diagnosed and classified using the same CNN technique | [31] |
InceptionV2, InceptionV3, MobileNet | 88.0 | Recommended for mobiles and embedded applications as MobileNet is light weight architecture and fast model | [30] |
CNN, VGG-16 model | 88.0 | The accuracy of the system can be improved by increasing the size of dataset and new deep neural network models can also be considered | [29] |
Image processing, SVM | 90.0 | The system can be extended for classifying other diseases | [28] |
CNN using TensorFlow | 75.2 | The system can be implemented in android device using Tensorflow lite | [27] |
Deep CNN in addition to GoogleNet | 94.9 | The model are able to detect images that do not belong to the eight used classes (classified as unknown images) | [22] |
Neural and fuzzy approach | 94.5 | The proposed method improves the performance by 4.9% | [23] |
Otsu algorithm, Alex and VGG-16 model | 99.0 | Better results were achieved compared to existing methods | [24] |
Deep CNN | 91.9 | The used model is more reliable and robust compared with existing transfer learning models | [25] |
CNN, Random Forest, KNN, Single-layered perceptron | 93.6-97.9 | The proposed method can perform several routine pathologist tasks | [26] |