基础
- cnn
Conv2D, Conv1D
filters, kernel_size, strides, padding
- pooling
MaxPooling, AveragePooling
tf.keras.layers.MaxPool2D
经典模型
参考ng课程: classic-networks
- LeNet
- AlexNet
- VGG-16
- ResNet
- Inception
- MobileNet
- EfficientNet
训练技巧
- transfer learning
- 参数初始化
- data augmentation
- 优化器
- 学习率变化
应用
-
objection dection
yolo模型 -
semantic segmentation
U-Net模型 -
Face recognition
One shot learning -
Neural style transfer
-
案例
TensorFlow images案例教学
案例包括:
1. cnn网络
2. 图像分类
3. Cransfer learning and fine-tuning
4. Data Augmentation
5. Image segmentation
6. Object detection
网友评论