MANet

RGBT234 dataset

链接：https://pan.baidu.com/s/1weaiBh0_yH2BQni5eTxHgg 提取码：qvsq

RGBT210 dataset

链接：https://pan.baidu.com/s/1FClmX0SH3WarcczkEQbmwA 提取码：ps8j

GTOT dataset

链接：https://pan.baidu.com/s/1zaR6aXh9PVQs063Q_b9zQg 提取码：ajma

RGBT234 toolkit

链接：https://pan.baidu.com/s/1UksOGtD2yl6k8mtB-Wr39A 提取码：4f68

RGBT210 toolkit

链接：https://pan.baidu.com/s/1KHMlbhu5R29CJvundGL4Sw 提取码：8wtc

GTOT toolkit

链接：https://pan.baidu.com/s/1iVVAXS4LZLvoQSGQnz7ROw 提取码：d53m

MANet result

MANet result in paper have upload in here, the report reslut is PR_0.777 SR_0.539 on RGBT234, PR_0.894 SR_0.724 on GTOT.

Multi-Adapter RGBT Tracking implementation on Pytorch

this code is update version based on submitted for VOT RGBT race code simplified version. So there are some differences from MANET's paper.

Prerequisites

CPU: Intel(R) Core(TM) i7-7700K CPU @ 3.75GHz GPU: NVIDIA GTX1080 Ubuntu 16.04

• python2.7
• pytorch == 0.3.1
• numpy
• PIL
• by yourself need install some library functions

Pretrained model for MANet

In our tracker, we use an VGG-M Net variant as our backbone, which is end-to-end trained for visual tracking.

The train on gtot model file in models folder,name called MANet-2IC.pth ,you can use this tracking rgbt234

Then,You need to modify the path in the tracking/options.py file depending on where the file is placed. It is best to use an absolute path. you can change code version of CPU/GPU in this flie

Train

you can use RGBT dataset as train data , in pretrain floder you need first genrate sequence list .pkl file use prepro_data.py , sencod change your data path , fainlly excute train.py

pretrain model :https://drive.google.com/open?id=1aO6LhOTxmpd7o_JXPLPjL3LsrQ5oqbl7

Run tracker

in the tracking/run_tracker.py file you need change dataset path and save result file dirpath in the tracking/options.py file you need set model file path ,and set learning rate depend on annotation. in tracking and train stage you need update modules/MANet3x1x1_IC.py file depend on annotation.

tracking model:https://drive.google.com/open?id=1Png508G4kQPI6HNewKQ4cfS36CvoSFSN

Result