/TrafficSign-ComputerVision-python

Predict traffic sign boards using Deep-Learning Model (Tensorflow)

Primary LanguagePythonMIT LicenseMIT

TrafficSign-ComputerVision-python

Python3 Top-down learning path: TrafficSign-ComputerVision-python GitHub stars GitHub forks GitHub Licence Linkedin Github VS Code

Description

Predict traffic sign boards using Deep-Learning Model (Tensorflow)

⚠️ Limitations

  • Due to my system limitaion I can't create very deep nerural networks
  • I can only taken batch_size = 32 (system limitaions)
  • Only trained neural networks by 2 epochs (system limitaions)

💡 Suggestions

  • Increase depth of nreural networks
  • Increase batch_size
  • Increase number of epochs
  • Experiment with preprocessor

Requirements

Programming Language

  • Python3 (VS Code)

Libraries

Data Cleaning
  • os
  • glob
  • shutil
  • csv
  • sklearn (train-val split)
Data Manipulation
  • Numpy
  • Pandas
DeepLearning Model
  • Tensorflow
  • Keras

Project Structure

  • Data Cleaning

DeepLearning Model

  • Sequential Layers
  • Created generators
  • Fitted model using Train Data
  • Saved DeepLearning Model in h5 file format
  • Load saved Model
  • Evalute Test data using using saved Model
  • predict Traffic-Sign boards using Model

Project Setup

  • Clone project
  • Create a branch
  • Make changes in Sequential Layers and preprocessor to reduce validation loss
  • Fit model by using Train Data
  • Save Model in h5 format
  • Create a pull request

License

CC0