三维项目-DeepVO的Pytorch实现-利用深度循环卷积神经网络实现端到端视觉里程计-优质项目实战.zip

# Readme - Model -  ## Usage - Download KITTI data and our pretrained model - This shell ```KITTI/downloader.sh``` can be used to download the KITTI images and pretrained model - the shell will only keep the left camera color images (image_03 folder) and delete other data - the downloaded images will be placed at ```KITTI/images/00/```, ```KITTI/images/01```, ... - the images offered by KITTI is already rectified - the direct [download link](https://drive.google.com/file/d/1l0s3rYWgN8bL0Fyofee8IhN-0knxJF22/view) of pretrained model - Download the ground truth pose from [KITTI Visual Odometry](http://www.cvlibs.net/datasets/kitti/eval_odometry.php) - you need to enter your email to request the pose data [here](http://www.cvlibs.net/download.php?file=data_odometry_poses.zip) - and place the ground truth pose at ```KITTI/pose_GT/``` - Run 'preprocess.py' to - remove unused images based on the readme file in KITTI devkit - convert the ground truth poses from KITTI (12 floats [R|t]) into 6 floats (euler angle + translation) - and save the transformed ground truth pose into ```.npy``` file - Pretrained weight of FlowNet ( CNN part ) can be downloaded [here](https://drive.google.com/drive/folders/0B5EC7HMbyk3CbjFPb0RuODI3NmM) - note that this pretrained FlowNet model assumes that RGB value range is [-0.5, 0.5] - the code of CNN layers is modified from [ClementPinard/FlowNetPytorch](https://github.com/ClementPinard/FlowNetPytorch) - Specify the paths and changes hyperparameters in ```params.py``` - If your computational resource is limited, please be careful with the following arguments: - ```batch_size```: choose batch size depends on your GPU memory - ```img_w```, ```img_h```: downsample the images to fit to the GPU memory - ```pin_mem```: accelerate the data excahnge between GPU and memory, if your RAM is not large enough, please set to False - Run ```main.py``` to train the model - the trained model and optimizer will be saved in ```models/``` - the records will be saved in ```records/``` - Run ```test.py``` to output predicted pose - output to ```result/``` - file name will be like ``out_00.txt`` - Run ```visualize.py``` to visualize the prediction of route - Other files: - ```model.py```: model is defined here - ```data_helper.py```: customized PyTorch dataset and sampler - the input images is loaded batch by batch ## Download trained model Provided by [alexart13](https://github.com/alexart13). - [trained model](https://drive.google.com/file/d/1l0s3rYWgN8bL0Fyofee8IhN-0knxJF22/view) - [optimizer](https://drive.google.com/file/d/1JlVJwEZy4W4EmgtTCNWmM4YAACUHxnr2/view) ## Required packages - pytorch 0.4.0 - torchvision 0.2.1 - numpy - pandas - pillow - matplotlib - glob ## Result - Training Sequences <table border=1> <tr> <td> <img src="https://user-images.githubusercontent.com/32840403/51480655-96989400-1da2-11e9-9cd7-c0618ed3ff6d.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480656-96989400-1da2-11e9-99d4-b55546e89d54.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480645-95676700-1da2-11e9-842e-ff20a243013d.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480647-95676700-1da2-11e9-95ac-d548d3aec24c.png" width="24%"/> </td> </tr> <tr> <td> <img src="https://user-images.githubusercontent.com/32840403/51480651-95fffd80-1da2-11e9-8b37-4cebfa7358e6.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480652-96989400-1da2-11e9-9096-7c93de6109da.png" width="24%"/> </td> </tr> </table> - Testing Sequence <table border=1> <tr> <td> <img src="https://user-images.githubusercontent.com/32840403/51480646-95676700-1da2-11e9-9723-ad8e07aa4706.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480648-95fffd80-1da2-11e9-917e-f9e67c1f1500.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480650-95fffd80-1da2-11e9-8c02-96aa37beab94.png" width="24%"/> <img src="https://user-images.githubusercontent.com/32840403/51480654-96989400-1da2-11e9-8202-ddedae1a9a7a.png" width="24%"/> </td> </tr> </table>