I have planned to create a deep learning model that classifies skin diseases(around 5 to 7 diseases). Please suggest me a good deep learning model to go with. I am planning to integrate this model as a mobile application. It goes like this. The user uploads a skin disease image and the machine is going to classify it accordingly. I need a good degree of accuracy.
If you are classifying images, you should obviously use CNN for good accuracy. The dataset you are classifying also plays a great role in determining the hyperparameters, optimizers, loss function.
So you should start with a basic CNN model that you can train with optimizers such as SGD and try tuning the hyperparameters accordingly.
If you are asking about the transfer learning models you can use RezNet or GoogLeNet
For the classification of images, use the Convolutional Neural network(CNN) Or you can opt for the Pre-trained model that was trained on a large amount of dataset Which provides good accuracy. Examples of the pre-trained models are VGG, MobileNet, Inception V3, etc.
For choosing various hyperparameters you can use the grid search approach.
There are tons and tons of architectures that you can try even in convolutional neural networks. If you have a sufficiently good amount of data, you can apply any of the ResNets (ResNet18, ResNet34, ResNet50 etc..), ResNEXTs, or any light image architecture is sufficiently good for such tasks, it gives a great balance between performance and time of execution. But if you want to achieve a really high score and have a sufficiently huge amount of data, you can refer to Efficient-Nets for your models.