Welcome to the CNN Models repository! This repo contains implementations of popular Convolutional Neural Network (CNN) architectures, designed for learning and experimentation purposes. All models will be deployed on Hugging Face for easy access and use.
You can check out the models here:
- AlexNet: TanishRajput/Alexnet
- VGG-16: TanishRajput/VGG-16
- Inception-v1 (GoogLeNet): TanishRajput/Inception-v1
- ResNet-50: TanishRajput/ResNet-50
- Inception-v3: TanishRajput/Inception-v3
More model links will be added soon!
Convolutional Neural Networks (CNNs) are a class of deep learning models particularly effective for image-related tasks such as classification, detection, and segmentation. This repository provides clean, well-documented implementations of famous CNN architectures built using TensorFlow/Keras.
| Model | Key Idea | Dataset | Performance |
|---|---|---|---|
| AlexNet | A pioneering CNN that popularized deep learning. | CIFAR-10 | Optimized for small image sizes. |
| VGG-16 | Deep network with uniform 3×3 convolution layers. | CIFAR-10 | Reached 94.2% accuracy with tuning. |
| Inception-v1 | Multi-scale feature extraction using Inception modules. | CIFAR-10 | Achieved 91.21% test accuracy with custom training. |
| ResNet-50 | Introduced residual connections to combat vanishing gradients. | CIFAR-100 | Achieved 87.16% test accuracy with transfer learning + data augmentation. |
| Inception-v3 | Enhanced Inception modules with factorized convolutions and auxiliary classifiers. | CIFAR-100 | Achieved 83.53% test accuracy with transfer learning. |
(More models coming soon!)
You can try the deployed models here:
-
AlexNet:
-
🌐 Streamlit App: Link
-
VGG-16:
-
🌐 Streamlit App: Link
-
Inception-v1 (GoogLeNet):
🌐 Streamlit App: Link -
ResNet-50:
🌐 Streamlit App: Link -
Inception-v3:
🌐 Streamlit App: Link
-
Clone the repo:
git clone https://github.com/tanishra/CNN-Models.git cd CNN-Models -
(Optional) Create and activate a virtual environment:
python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate
-
Install dependencies for a specific model:
cd alexnet pip install -r requirements.txtcd vgg-16 pip install -r requirements.txtcd Inception-v1 pip install -r requirements.txtcd ResNet-50 pip install -r requirements.txtcd Inception-v3 pip install -r requirements.txt -
Run the model scripts or Streamlit apps according to the folder you choose.
streamlit run app.py
- Each model folder contains:
- Model code: Implementation of the CNN architecture.
- Training script: Code to train the model on datasets like CIFAR-10, CIFAR-100, etc.
- Evaluation script: Test the trained model on validation or test sets.
Contributions are welcome! Feel free to:
- Add more CNN architectures.
- Improve existing implementations.
- Add tutorials or notebooks demonstrating usage. Please fork the repo and create a pull request.