This project implements a Convolutional Neural Network (CNN) from scratch using TensorFlow/Keras to classify traffic signs. The goal is to build a robust image classification pipeline that can accurately identify various traffic signs from images.
This project demonstrates:
- Preprocessing image data for training deep learning models.
- Building a custom CNN architecture using TensorFlow/Keras.
- Training the model and evaluating its accuracy.
- Saving and loading model checkpoints for later use.
Glottal Closure Instants (GCIs) correspond to the temporal locations of significant excitation to the vocal tract occurring during the production of voiced speech. GCI detection from speech signals is a well-studied problem given its importance in speech processing. Most of the existing approaches for GCI detection adopt a two-stage approach (i) Transformation of speech signal into a representative signal where GCIs are localized better, (ii) extraction of GCIs using the representative signal obtained in first stage. The former stage is accomplished using signal processing techniques based on the principles of speech production and the latter with heuristic-algorithms such as dynamic-programming and peak-picking. These methods are thus task-specific and rely on the methods used for representative signal extraction. However, in this paper, we formulate the GCI detection problem from a representation learning perspective where appropriate representation is implicitly learned from the raw-speech data samples. Specifically, GCI detection is cast as a supervised multi-task learning problem solved using a deep convolutional neural network jointly optimizing a classification and regression cost. The learning capability is demonstrated with several experiments on standard datasets. The results compare well with the state-of-the-art algorithms while performing better in the case of presence of real-world non-stationary noise.
The dataset contains labeled images of traffic signs. Each image represents one of several possible traffic sign categories.
- Number of classes:
43(if using German Traffic Sign Recognition Benchmark - GTSRB) - Image size: 32x32 RGB
- Dataset source (if GTSRB): Kaggle - GTSRB
You can place the dataset in the data/ folder or write a script to download and extract it automatically.
The CNN model is composed of:
- 3 Convolutional Layers with ReLU activations
- MaxPooling after each convolution
- Flatten layer
- 2 Dense layers including final Softmax output
The model is compiled with:
- Loss:
categorical_crossentropy - Optimizer:
Adam - Metrics:
accuracy
Clone the repo and install required dependencies:
git clone https://github.com/YOUR_USERNAME/Traffic-Sign-CNN.git
cd Traffic-Sign-CNN
pip install -r requirements.txt