lukemelas/EfficientNet-PyTorch
EfficientNet PyTorch
This is a PyTorch implementation of EfficientNet convolutional neural networks. It serves as a computer vision model library providing architectures for image classification and high-level feature extraction, including pre-trained weights for immediate image categorization.
The library supports transfer learning by allowing the modification of model architectures and output layers to accommodate a custom number of classes for new datasets. It also includes a model exporter to convert trained PyTorch weights into the ONNX format for production inference.
The system covers broader computer vision capabilities, such as extracting intermediate feature maps and loading both standard and adversarial pre-trained model weights.
Features
- Image Classification - Provides scaled convolutional neural network architectures for categorizing images into predefined groups.
- Computer Vision Models - Offers a comprehensive library of scalable neural network architectures for vision tasks.
- Depthwise Separable Convolutions - Uses depthwise-separable convolutions to reduce computational costs while maintaining performance.
- EfficientNet Implementations - Provides a full PyTorch implementation of EfficientNet convolutional neural networks.
- Feature Extraction - Enables extraction of high-level image representations and intermediate feature maps.
- Convolutional Classifiers - Provides pre-trained convolutional classifiers for immediate image categorization.
- Compound Scaling - Implements the compound scaling coefficient to balance network depth, width, and resolution.
- Vision Model Weight Loading - Supports importing pre-trained weights, including standard and adversarial versions, for vision tasks.
- Transfer Learning - Adapts pre-trained models for custom datasets by modifying the output layers.
- Classifier Head Management - Provides tools for modifying classification heads to accommodate custom dataset class counts.
- Squeeze-and-Excitation Extractors - Includes squeeze-and-excitation blocks to recalibrate channel-wise feature responses.
- Bottleneck Layers - Utilizes inverted bottleneck layers to maintain information flow with fewer parameters.
- ONNX Model Exporters - Supports converting trained PyTorch weights into the ONNX format for production deployment.
- PyTorch Model Export - Converts trained PyTorch neural networks into ONNX format for production inference.
- رؤية الحاسوب - Efficient classification models.
- Neural Network Architectures - EfficientNet architecture implementation.
- أدوات المطور - EfficientNet implementation.
- Plug and Play Modules - Model scaling modules for efficient network design.
- Architecture Search - Listed in the “Architecture Search” section of the The Incredible Pytorch awesome list.
- CNN - Listed in the “CNN” section of the Ailia Models awesome list.
