NgNet is a car detector based on KittiBox.
Statistics generated from running 'evaluate.py' on VGG, Resnet50, and Resnet100 models.
| train easy | train moderate | train hard | val easy | val moderate | val hard | speed (msec) | speed (fps) | post (msec) | |
|---|---|---|---|---|---|---|---|---|---|
| VGG | 93.25% | 84.10% | 69.09% | 94.27% | 86.32% | 70.78% | 104.4554 | 9.5735 | 3.5804 |
| Resnet50 | 98.46% | 91.50% | 79.41% | 96.78% | 86.47% | 72.33% | 59.9431 | 16.6825 | 3.1012 |
| Resnet100 | 98.56% | 93.58% | 81.19% | 96.01% | 89.13% | 75.02% | 98.4383 | 10.1506 | 2.9213 |
Difference:
| train easy | train moderate | train hard | val easy | val moderate | val hard | speed | |
|---|---|---|---|---|---|---|---|
| Resnet50 vs VGG | 5.21% | 7.4% | 10.35% | 2.51% | 0.15% | 1.55% | x1.74 |
| Resnet100 vs VGG | 5.31% | 9.48% | 12.1% | 1.74% | 2.81% | 4.24% | x1.06 |
The code requires Tensorflow 1.0 as well as the following python libraries:
- matplotlib
- numpy
- Pillow
- scipy
- runcython
- imageio
- opencv
Those modules can be installed using: pip install -r requirements.txt.
Read KittiBox README for detailed installation.
Generate data from Udacity CrowdAI and AUTTI using my version of vod-converter which is compatible with both Python 2 and 3.
Note: this converter is a fork from umautobots vod-converter and it contains some changes to make it work for this repositoty.
This project started out as a fork of KittiBox.
Data convertion tool is a fork from umautobots's vod-converter but has some minnor changes to work with Python 2 and 3.