Merge branch 'master' of github.com:tom-roddick/oft

readme.md

@@ -1,8 +1,14 @@
 # Orthographic Feature Transform for Monocular 3D Object Detection
 
 ![OFTNet-Architecture](https://github.com/tom-roddick/oft/raw/master/architecture.png "OFTNet-Architecture")
-This is a PyTorch implementation of the OFTNet network from the paper [Orthographic Feature Transform for Monocular 3D Object Detection](https://arxiv.org/abs/1811.08188). The code currently supports training the network from scratch on the KITTI dataset - intermediate results can be visualised using Tensorboard. The current version of the code is intended primarily as a reference, and for now does not support decoding the network outputs into bounding boxes via non-maximum suppression. This will be added in a future update. Note also that there are some slight implementation differences from the original code used in the paper. Please see `train.py` for details of training options.
+This is a PyTorch implementation of the OFTNet network from the paper [Orthographic Feature Transform for Monocular 3D Object Detection](https://arxiv.org/abs/1811.08188). The code currently supports training the network from scratch on the KITTI dataset - intermediate results can be visualised using Tensorboard. The current version of the code is intended primarily as a reference, and for now does not support decoding the network outputs into bounding boxes via non-maximum suppression. This will be added in a future update. Note also that there are some slight implementation differences from the original code used in the paper.
 
+## Usage
+The training script can be run by calling `train.py` with the name of the experiment as a required position argument. 
+```
+python train.py name-of-experiment --gpu 0
+```
+By default data will be read from `data/kitti/objects` and model checkpoints will be saved to `experiments`. The model is trained using the KITTI 3D object detection benchmark which can be downloaded from [here](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d). See `train.py` for a full list of training options.
 
 ## Citation
 If you find this work useful please cite the paper using the citation below.
@@ -13,4 +19,4 @@ If you find this work useful please cite the paper using the citation below.
   journal={British Machine Vision Conference},  
   year={2019}  
 }
-```
+```