RT-MDNet-master
所属分类：其他
开发工具：Python
文件大小：115KB
下载次数：0
上传日期：2020-07-31 15:09:36
上 传 者：
木子231
说明： 一种新颖的视觉跟踪算法，基于一个来自有区别训练的CNN的表示。算法中使用大量跟踪视频的ground-truth来对CNN进行预训练，以获得通用的目标表示。
(A novel visual tracking algorithm based on a representation from CNN with differentiated training. In the algorithm, a large number of ground truth tracking videos are used to pre train CNN to obtain a common target representation.)
文件列表:
Run.py (5752, 2020-07-21)
__init__.py (0, 2020-07-21)
modules (0, 2020-07-21)
modules\__init__.py (0, 2020-07-21)
modules\bbreg.py (2036, 2020-07-21)
modules\data_prov.py (7198, 2020-07-21)
modules\img_cropper.py (2789, 2020-07-21)
modules\model.py (6735, 2020-07-21)
modules\prepro_data.py (3022, 2020-07-21)
modules\prepro_data_imagenet.py (2698, 2020-07-21)
modules\pretrain_options.py (968, 2020-07-21)
modules\roi_align (0, 2020-07-21)
modules\roi_align\Makefile (157, 2020-07-21)
modules\roi_align\__init__.py (0, 2020-07-21)
modules\roi_align\_ext (0, 2020-07-21)
modules\roi_align\_ext\__init__.py (0, 2020-07-21)
modules\roi_align\_ext\roi_align (0, 2020-07-21)
modules\roi_align\_ext\roi_align\__init__.py (383, 2020-07-21)
modules\roi_align\_ext\roi_align\_roi_align.so (143391, 2020-07-21)
modules\roi_align\build.py (833, 2020-07-21)
modules\roi_align\functions (0, 2020-07-21)
modules\roi_align\functions\__init__.py (0, 2020-07-21)
modules\roi_align\functions\roi_align.py (5193, 2020-07-21)
modules\roi_align\modules (0, 2020-07-21)
modules\roi_align\modules\__init__.py (0, 2020-07-21)
modules\roi_align\modules\roi_align.py (2861, 2020-07-21)
modules\roi_align\src (0, 2020-07-21)
modules\roi_align\src\cuda (0, 2020-07-21)
modules\roi_align\src\cuda\Makefile (170, 2020-07-21)
modules\roi_align\src\cuda\roi_align.cu.o (70760, 2020-07-21)
modules\roi_align\src\cuda\roi_align_kernel.cu (22737, 2020-07-21)
modules\roi_align\src\cuda\roi_align_kernel.h (3576, 2020-07-21)
modules\roi_align\src\roi_align_cuda.c (7768, 2020-07-21)
modules\roi_align\src\roi_align_cuda.h (1138, 2020-07-21)
modules\sample_generator.py (3270, 2020-07-21)
modules\utils.py (3194, 2020-07-21)
options.py (1293, 2020-07-21)
... ...
## RT-MDNet: Real-Time Multi-Domain Convolutional Neural Network Tracker
Created by [Ilchae Jung](http://cvlab.postech.ac.kr/~chey0313), [Jeany Son](http://cvlab.postech.ac.kr/~jeany), [Mooyeol Baek](http://cvlab.postech.ac.kr/~mooyeol), and [Bohyung Han](http://cvlab.snu.ac.kr/~bhhan)
### Introduction
RT-MDNet is the real-time extension of [MDNet](http://cvlab.postech.ac.kr/research/mdnet/) and is the state-of-the-art real-time tracker.
Detailed description of the system is provided by our [project page](http://cvlab.postech.ac.kr/~chey0313/real_time_mdnet/) and [paper](https://arxiv.org/pdf/1808.08834.pdf)
### Citation
If you're using this code in a publication, please cite our paper.
@InProceedings{rtmdnet,
author = {Jung, Ilchae and Son, Jeany and Baek, Mooyeol and Han, Bohyung},
title = {Real-Time MDNet},
booktitle = {European Conference on Computer Vision (ECCV)},
month = {Sept},
year = {2018}
}
### System Requirements
This code is tested on *** bit Linux (Ubuntu 16.04 LTS).
**Prerequisites**
0. PyTorch (>= 0.2.1)
0. For GPU support, a GPU (~2GB memory for test) and CUDA toolkit.
0. Training Dataset (ImageNet-Vid) if needed.
### Online Tracking
**Pretrained Model and results**
If you only run the tracker, you can use the pretrained model:
[RT-MDNet-ImageNet-pretrained](https://www.dropbox.com/s/lr8uft05zlo21an/rt-mdnet.pth?dl=0).
Also, results from pretrained model are provided in [here](https://www.dropbox.com/s/pefp4dqjwjows3z/RT-MDNet%20Results.zip?dl=0).
**Demo**
0. Run 'Run.py'.
### Learning RT-MDNet
**Preparing Datasets**
0. If you download ImageNet-Vid dataset, you run 'modules/prepro_data_imagenet.py' to parse meta-data from dataset. After that, 'imagenet_refine.pkl' is generized.
0. type the path of 'imagenet_refine.pkl' in 'train_mrcnn.py'
**Demo**
0. Run 'train_mrcnn.py' after hyper-parameter tuning suitable to the capacity of your system.
近期下载者：
相关文件：
收藏者：