DIP big job
Age Recognition with Deep Learning and CNN
Deep Learning-Based Approaches
The basic steps | deep learning method |
---|---|
1.(10%) | Requirements analysis system design; environment construction |
2.(10%) | Dataset and preprocessing |
3.(40%) | Convolutional neural network model design; model programming; model training; model testing |
4.(30%) | Experimental structural analysis: (1) Longitudinal comparison. Model training and comparison of different parameters (such as learning rate, number of iterations, batchsize), different structures (such as different layers, different numbers of convolution kernels, etc.), different activation functions, and different i gradient descent methods of the network built by yourself ; (2) Horizontal comparison. Perform a performance comparison of networks with different architectures. (3) Visualize the relationship between the model loss and the number of iterations, accuracy, and number of iterations and other values through tensorboard. (4) The face recognition project also needs to compare the accuracy of different experimental schemes (such as lighting, expression, occlusion area, etc.) ) analysis results. |
5(10%) | System Integration and Improvement |
Deep Learning = Data + Automatic Feature Learning Model
Process Synthesis