This project implements a deep learning pipeline for car image classification using ResNet-based Convolutional Neural Networks (CNNs).
The model is built with PyTorch and leverages transfer learning from pretrained ImageNet weights (resnet18 or resnet34).

The repository also includes a web application that allows interactive training and inference directly from the browser.

• CNN classifier built on ResNet18 / ResNet34 architectures.
• Transfer learning with optional pretrained weights.
• Configurable number of frozen layers for fine-tuning.
• Training and inference modes accessible via command-line or web interface.
• Logging of training metrics and model checkpoints.

git clone https://github.com/jacons/cnn_webapp
conda env create -f environment.yml
conda activate car-classifier

The dataset can be downloaded from Google Drive: Car Dataset (Google Drive Link) After downloading, extract it and place it under the datasets/ folder (e.g., datasets/car_dataset/).

The application supports both training and inference modes.

Training Mode

python app.py --train

Inference Mode

python app.py --inference

You can also train the model directly from the terminal with full configuration options.

Example:

python car_classifier.py \
    --dataset_path "datasets/car_dataset" \
    --metric_history "results/r18_pretrained_fre0.json" \
    --new_hist \
    --model_cache "results/r18_pretrained_fre0.pth" \
    --batch_size 150 \
    --model_name "resnet18" \
    --pretrained \
    --freeze_layers 0 \
    --num_epochs 15 \
    --device "cuda:2"

Below are examples of the training dashboard and inference interface:

🔹 Training Interface

🔹 Inference Interface

MIT

• @jacons

• Code is written entirely by the author.
• Documentation is generated by ChatGPT and later reviewed and validated by the author.