Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/alessiamarcolini/deepstreet
Traffic Sign Recognition - Fine tuning VGG16 + GTSRB
https://github.com/alessiamarcolini/deepstreet
convolutional-neural-networks deeplearning keras traffic-sign-classification
Last synced: about 20 hours ago
JSON representation
Traffic Sign Recognition - Fine tuning VGG16 + GTSRB
- Host: GitHub
- URL: https://github.com/alessiamarcolini/deepstreet
- Owner: alessiamarcolini
- License: mit
- Created: 2017-07-07T09:00:53.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2022-08-10T12:45:52.000Z (over 2 years ago)
- Last Synced: 2023-10-20T16:22:09.263Z (about 1 year ago)
- Topics: convolutional-neural-networks, deeplearning, keras, traffic-sign-classification
- Language: Python
- Size: 488 KB
- Stars: 67
- Watchers: 5
- Forks: 23
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE.txt
Awesome Lists containing this project
README
# Deepstreet
> Deepstreet is the project I developed for my high school thesis in IT @ ITI Marconi, Verona (IT).
This project aims to provide a system able to recognize the type of a street sign into an image, using Deep Learning techniques.
## Features
- Written in Python
- Keras as a main library for Deep Learning, with Numpy and OpenCV
## Getting Started
### Prerequisistes
To run all the scripts you need the following packages:
- Python version 3.5
- `numpy` v. 1.13
- `matplotlib` v. 2.0
- `OpenCV` v. 3.0
- `tensorflow` v. 1.1 -- or `tensorflow-gpu` if you have GPUs on your machine
- `keras` v. 2.0
- `hdf5` v. 1.8 and `h5py` v. 2.7
Optional, but recommended:
- **NVIDIA cuDNN** if you have NVIDIA GPUs on your machine.
[https://developer.nvidia.com/rdp/cudnn-download]()
The easiest way to get (most) these is to use an all-in-one installer such as [Anaconda](http://www.continuum.io/downloads) from Continuum. This is available for multiple architectures.
### Dataset
Download the **German Traffic Sign Recognition Benchmark (GTSRB)** from https://benchmark.ini.rub.de/
### Running scripts
To run the scripts, just download this repo and execute:
```shell
python
```
The two main scripts (`deepstreet_training.py` and `deepstreet_predict.py`) can be executed with optional arguments. To find out the arguments for, let's say, `deepstreet_training.py` execute:
```shell
python deepstreet_training.py --help
```
```shell
Using TensorFlow backend.
usage: deepstreet_training.py [-h] [--gpu GPU] [--epochs EPOCHS]
[--output_dir OUTPUT_DIR]
[--input_dir INPUT_DIR] [--debug DEBUG]
Run a training experiment using pretrained VGG16, specified on the deepstreet
DataSet.
optional arguments:
-h, --help show this help message and exit
--gpu GPU GPU Device (default: 0)
--epochs EPOCHS Number of Epochs during training (default: 10)
--output_dir OUTPUT_DIR
Output directory
--input_dir INPUT_DIR
Input directory
--debug DEBUG Debug mode
```
## Contributing Guidelines
All contributions and suggestions are welcome!
For suggested improvements, please create an issue.
## License
This project is licensed under the MIT License - see the [LICENSE.txt](https://github.com/alessiamarcolini/deepstreet/blob/master/LICENSE.txt) file for details.