A web-based application that uses a pre-trained machine learning model to classify uploaded traffic sign images and display the predicted traffic sign label with high accuracy.

To design and deploy a deep learning model that can accurately recognize and classify traffic signs from images in real time.

Python, TensorFlow/Keras, Django, React, HTML, CSS, JavaScript.

One of the major challenges during development was training a model that generalized well without overfitting or underfitting. The model also struggled to distinguish between speed limit signs with similar shapes but different numerical values (e.g., 20 km/h and 50 km/h), resulting in incorrect predictions.

To improve performance, I refined the training process by increasing the diversity of the training data through data augmentation, tuning the model's hyperparameters, and training for an optimal number of epochs. I also improved image preprocessing to ensure clearer feature extraction. These changes reduced misclassifications and improved the model's ability to correctly recognize numerical traffic signs.