blitu12345 / traffic-signs

Using CNN to detect inner representation of 43 different Traffic Signs which is provided in GTRSB dataset

Recognising Traffic Signs using Convolutional Neural Network

---

Recognise the traffic signs from the images into 43 different category.

Model:

• 3 Convolutional block with 32, 64 and 128 filters of 3x3 kernel size
• Activation function used ReLu
• Optimiser used Stochastic Gradient Descent (SGD)
• Loss is Categorical Crossentropy

Dataset:

• Link: GTRSB
• 43 classes
• More than 50,000 images in total
• Large, lifelike database
• Physical traffic sign instances are unique within the dataset