The image recognition is basically done with the Convolutional Neural Networks(CNN’s), now it can be done through TensorFlow in an efficient way. TensorFlow is the machine learning Framework preferred by most of the researchers as it finds space in handling more efficient and complex tasks easily.
The convolution involves, transforming input image to matrices with different layers of dimensions. These are the Processes involved in the whole Process of image recognition in Neural Networks with TensorFlow,
(I) Feature Learning,
- Convolution,
- RELU – Activation Function
- Pooling
(II) Classification
- Fully Connected Layer
- Softmax Classification Function
For our Project( WandResearch ) I preferred to use Convolutional Neural Network(CNN) as a starter for image recognition purposes.
MLNOTES 