文件列表:

efficientnet-b7.ipynb

# Arabic Letters Classification using EfficientNet_B7 This repository contains code for training an Arabic letters classification model using the EfficientNet_B7 architecture. The model achieves an impressive test accuracy of 97.04%. The dataset used for training is the [Final Arabic Alpha dataset](https://www.kaggle.com/competitions/arabic-letters-classification/data), and the code is implemented in PyTorch. ## Dataset The dataset consists of Arabic letters organized into folders, with each folder representing a different class. The `MakeDataset` class is implemented to load and preprocess the dataset using PyTorch's DataLoader. ## Data Preprocessing The images are preprocessed using the following transformations: - Conversion to RGB format - Grayscale conversion - Resizing to (224, 224) pixels - Normalization with mean [0.485, 0.456, 0.406] and standard deviation [0.229, 0.224, 0.225] ## Model Architecture The EfficientNet_B7 model is utilized for feature extraction. The classifier's last layer is replaced with a linear layer with the correct number of output classes (65 in this case). ## Training The model is trained using the Adam optimizer with a learning rate of 0.0001. The learning rate is adjusted using a step-wise scheduler. The training loop is designed to monitor and display training and validation losses, as well as accuracy during each epoch. ## Evaluation The model is evaluated on a test set, and the final test accuracy is calculated. Additionally, loss and accuracy curves are plotted over the training epochs. ## Results The trained model achieves a test accuracy of 97.04%. The loss and accuracy curves provide insights into the model's training and validation performance. ## Confusion Matrix A confusion matrix is generated using the predictions on the test set, providing a detailed view of the model's classification performance. ## Usage To train the model, run the provided notebook. Make sure to adjust hyperparameters and paths as needed. ## Dependencies - PyTorch - torchvision - scikit-learn - matplotlib - PIL ## Acknowledgments - The EfficientNet_B7 model is from the torchvision library. Feel free to use and modify the code for your own projects!

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：