【翻譯】TensorFlow卷積神經網絡識別CIFAR 10Convolutional Neural Network (CNN)| CIFAR 10 TensorFlow
原網址:https://data-flair.training/blogs/cnn-tensorflow-cifar-10/
by
2、卷積神經網絡
在開始之前我們要充分理解CNN的原理。通常我們采用CIFAR 10數據集,這是一個物體識別的數據集,由60000張32*32的圖片,包含10中類型的物體組成。
https://data-flair.training/blogs/wp-content/uploads/sites/2/2018/05/Typical_cnn.png
下載地址 https://www.cs.toronto.edu/~kriz/cifar.html.
3、CIFAR 10模型結構
該模型86%的正確率,需要在GPU上訓練幾小時。你需要下列文件:
cifar10_input.py Reads the native CIFAR-10 binary file format.
cifar10.py Builds the CIFAR-10 model.
cifar10_train.py Trains a CIFAR-10 model on a CPU or GPU.
cifar10_multi_gpu_train.py Trains a CIFAR-10 model on multiple GPUs.
cifar10_eval.py Evaluates the predictive performance of a CIFAR-10 model.
a.輸入
下面還沒翻譯完 稍等
【翻譯】TensorFlow卷積神經網絡識別CIFAR 10Convolutional Neural Network (CNN)| CIFAR 10 TensorFlow