Question

Question about deep learning:

1. What should I do for the preprocessing of a multi-label classification & detection algorithm?

2. What is the commonly used pre-trained model for the use of multi-label classification & detection?

Answer 1

Answer 1)

In multi-label classification, there are more than one labels which are associated with each sample in the data. Each sample has only one label associated with it.

For example, a given problem to classify a fruit like apple, orange, banana. Here there are three labels in the problem. In the preprocessing of this data set, we can convert these labels to 0, 1, 2 for the computer to understand them. Here what we have done is that we have assigned apple as 0, orange as 1, banana as 2. In this way, the preprocessing of the multi-label classification is done.

  

Answer 2) Pre-trained model for the use of multi-label classification

1) VGG - VGG is a convolutional neural network model for object-recognition in the Image. It has achieved very good performance on the ImageNet dataset which is the largest database of image over 14 million images.

2) Mask R-CNN-   Mask R-CNN is a flexible model developed for the purpose of object instance segmentation.

3) Inception: Also used for image recognition.