Is this directory still empty? When can you update？

ImportError: cannot import name region @

I evaluated the official checkpoint siamrpn_r50_l234_dwxcorr on VOT2018, I found that if i drop out bounding box smoothness , it only got about 0.36 EAO (fast away from 0.41 using smoothness). It shows that this post-processing refinement is important to boost the performance. But, why it was not mentioned at all in the paper?

Hi, as I followed the instruction of install.md, I encountered the c1.exe failed with exit status 2 problem when I run python setup.py build_ext --inplace. How can I fix it?
Thanks a lot.

Hi, I want to train the alexnet with your training code on VID dataset, is there any backbone use the same structure as SiamRPN++? Means random shift and depth-wise cross correlation used even in alexnet.

请问如何不通过命令行运行train.py?目前我直接在运行test.py 可以，但是直接运行train.py:
File "/home/public/anaconda3/lib/python3.6/os.py", line 669, in getitem
raise KeyError(key) from None
KeyError: 'RANK'

请问如何在代码中设置：
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
python -m torch.distributed.launch
--nproc_per_node=8
--master_port=2333
../../tools/train.py --cfg config.yaml
中的值，使得可以直接在pycharm运行train.py，
谢谢您的回复。

Would you mind, if possible, indicate in pysot/MODEL_ZOO.md the training data sources of each reported model for fair comparison purpose? Thanks a lot in advance.

can you offer the baidu wangpan for training dataset ?
thanks

self.search_cls_conv = nn.Conv2d(feature_in,
feature_in * cls_output, kernel_size=3)
self.search_loc_conv = nn.Conv2d(feature_in,
feature_in * loc_output, kernel_size=3)
should be :
self.search_cls_conv = nn.Conv2d(feature_in,
feature_in, kernel_size=3)
self.search_loc_conv = nn.Conv2d(feature_in,
feature_in, kernel_size=3)

Thank you for your work, Could I run train.py in single GPU?what should I do? When I run 'train.py' directly, I get the following error:Traceback (most recent call last):
File "../../tools/train.py", line 314, in
main()
File "../../tools/train.py", line 256, in main
rank, world_size = dist_init()
File "/home/db/Subject/pysot/pysot/utils/distributed.py", line 104, in dist_init
rank, world_size = _dist_init()
File "/home/db/Subject/pysot/pysot/utils/distributed.py", line 83, in _dist_init
rank = int(os.environ['RANK'])
File "/home/db/anaconda3/envs/pysot/lib/python3.7/os.py", line 678, in getitem
raise KeyError(key) from None
KeyError: 'RANK'

conda list:

cython　　　0.29.7　　　py37he6710b0_0　　　defaults

我执行：sudo python setup.py build_ext --inplace
出现错误：
Traceback (most recent call last):
File "setup.py", line 3, in
from Cython.Build import cythonize
ImportError: No module named 'Cython'

我明明已经装了Cython了，为什么还报错呢？

Hi, I'm wondering why it is named "SiamRPN", where I think it's more like "SiamSSD" since it is single staged. Just curious.

eval.py载入多个算法时
trackers = glob(os.path.join(args.tracker_path, args.dataset, args.tracker_prefix+'*'))
trackers = [x.split('/')[-1] for x in trackers]
这样语法会存在错误吧，是不是应该改成
trackers = args.tracker_prefix.split(" ")

Since you've stated that you only tested on Linux I wanted to let you know that your code works flawless for me on Windows as well.
https://github.com/STVIR/pysot/blob/master/INSTALL.md#installation

Thanks for sharing!

Excuse me, can I run under @the window system?

/home/public/anaconda3/bin/python -u ../../tools/test.py --snapshot model.pth --dataset VOT2018 --config config.yaml

but:ModuleNotFoundError: No module named 'pysot'

I have
export PYTHONPATH=/home/cbl/PycharmProjects/pysot-master/pysot:$PYTHONPATH

Hi,
Is there a way to use the pretrained baseline models as base for the training, rather than just the pretrained backbones?
I want to train a model on a new dataset and don't want to retrain it with the provided four.

Dear authors!
the pre-trained baseline models were trained by VID and youtube-bb dataset? or just VID is enough? I trained on VID and filter some large objects but the model overfits a lot. Any training details i will appreciate so much

to Download testing datasets ,
how to set the dataset?
could you say the detail of it ?
thanks

May I ask how to solve this error when running the demo.py.
File "/home/shiyuanyuan/anaconda3/envs/pysot/lib/python3.7/site-packages/torch/nn/modules/module.py", line 719, in load_state_dict
self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for ModelBuilder:
Missing key(s) in state_dict:

why I run it and false
conda create --name pysot python=3.7
conda activate pysot

CondaHTTPError: HTTP 404 NOT FOUND for url https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/noarch/repodata.json
Elapsed: 00:09.424438

The remote server could not find the noarch directory for the
requested channel with url: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge

As of conda 4.3, a valid channel must contain a noarch/repodata.json and
associated noarch/repodata.json.bz2 file, even if noarch/repodata.json is
empty. please request that the channel administrator create
noarch/repodata.json and associated noarch/repodata.json.bz2 files.
$ mkdir noarch
$ echo '{}' > noarch/repodata.json
$ bzip2 -k noarch/repodata.json

You will need to adjust your conda configuration to proceed.
Use conda config --show channels to view your configuration's current state.
Further configuration help can be found at https://conda.io/docs/config.html.

RT. Can you make some explanations, thanks.

The following error is encountered during (file pysot/testing_dataset/GOT-10k/GOT-10k_new.json, line 60)
with open(os.path.join(dataset_root, name+'_new.json'), 'r') as f:
This is due to the missing file GOT-10k_new.json which can neither be found on BaiduYun disk nor Google Drive. Would you mind upload this missing file? Thanks in advance.

Hi;
I want to save the model graph, but it occur errors when I save it via tb_writer.add_graph.
it seems like this issue： lanpa/tensorboardX#220

who can help me to visual network structure?

您好!
非常感谢你们开源工作,不知道能否提供百度云链接下载;国内谷歌云下载很不稳定,经常下载失败。

when use one node
File "/home/public/anaconda3/lib/python3.6/os.py", line 669, in getitem
raise KeyError(key) from None
KeyError: 'RANK'

I want to test trackers on VOT2019 dataset. Can you tell me how to write a json file for the new dataset?

从评估的baselines看，pysot是不是只提供了SiamRPN++和SiamMask加上几种backbone的模型，其他三种DaSiamRPN、SiamRPN、SiamFC没有提供，没找到可下载的链接。

how the set config.yaml?
and other operation

importError: cannot import name 'region'，总是出现该错误，使用pip install region也显示安装成功了，我想咨询下这个是什么原因

when do you offer the training code ?
thanks

你好。非常感谢提供的开源资源。在配置好环境后,运行python bin/eval.py报错。pyxbld/temp.linux-x86_64-3.7/pyrex/pysot/utils/region.c:594:24: fatal error: src/region.h: No such file or directory
#include "src/region.h"
现在找不到错误的原因，可否提供帮助

RuntimeError: PyTorch was compiled without NumPy support
这个怎么解决呀？得用什么版本的numpy呀

Hi, you said we should download models in PySOT Model Zoo and put the model.pth in the correct directory in experiments
But When I clicked the url link of siamrpn_r50_l234_dwxcorr on https://github.com/STVIR/pysot/blob/master/MODEL_ZOO.md, the web went to https://drive.google.com/open?id=1Q4-1563iPwV6wSf_lBHDj5CPFiGSlEPG which I can not open

TRAIN.md
Testing
python -u ../tools/test.py
--snapshot {}
--config config.py \

should be:
python -u ../../tools/test.py
--snapshot {}
--config config.yaml \

Thanks for your work.

I used the following command under ''/experiments/siammask_r50_l3/'':

CUDA_VISIBLE_DEVICES=0 # change your cuda visible devices here
python -m torch.distributed.launch --nproc_per_node=1 --master_port=2333 ../../tools/train.py --cfg config.yaml

But I encountered an error when I trained siammask:

Traceback (most recent call last):
  File "../../tools/train.py", line 316, in <module>
    main()
  File "../../tools/train.py", line 311, in main
    train(train_loader, dist_model, optimizer, lr_scheduler, tb_writer)
  File "../../tools/train.py", line 211, in train
    outputs = model(data)
  File "/data/anaconda3/envs/tang0.4.1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
  File "/data/anaconda3/envs/tang0.4.1/lib/python3.7/site-packages/pysot/utils/distributed.py", line 43, in forward
    return self.module(*args, **kwargs)
  File "/data/anaconda3/envs/tang0.4.1/lib/python3.7/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
  File "/data/anaconda3/envs/tang0.4.1/lib/python3.7/site-packages/pysot/models/model_builder.py", line 113, in forward
    outputs['total_loss'] += cfg.TRAIN.MASK_WEIGHT * mask_loss
TypeError: unsupported operand type(s) for *: 'int' and 'NoneType'

Could you please tell me how to run train.py correctly?

Thank you very much!

请问如何不通过命令行运行train.py?目前我直接在运行test.py 可以，但是直接运行train.py:
File "/home/public/anaconda3/lib/python3.6/os.py", line 669, in getitem
raise KeyError(key) from None
KeyError: 'RANK'

请问如何在代码中设置：
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
python -m torch.distributed.launch
--nproc_per_node=8
--master_port=2333
../../tools/train.py --cfg config.yaml
中的值，使得可以直接在pycharm运行train.py，
谢谢您的回复。

Traceback (most recent call last):
File "../../tools/train.py", line 378, in
main()
File "../../tools/train.py", line 327, in main
os.makedirs(cfg.TRAIN.LOG_DIR)
File "/home/public/anaconda3/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
PermissionError: [Errno 13] Permission denied: './logs'

there is no cfg.TRAIN.LOG_DIR in the config.yaml
why os.makedirs(cfg.TRAIN.LOG_DIR)?

is it ok for cuda 9.0 9.1 9.2 ,torch0.4.0?

When processing training dataset, how to generate corresponding annotations after cropping the training dataset?

Hi,
when running the training for siamrpn_alex_dwxcorr as described, I always run into an error that seems to be related to tensor shapes:

Traceback (most recent call last):
File "../../tools/train.py", line 316, in
main()
File "../../tools/train.py", line 311, in main
train(train_loader, dist_model, optimizer, lr_scheduler, tb_writer)
File "../../tools/train.py", line 211, in train
outputs = model(data)
File "/home/users/kob/mastera/pysot/venv/lib/python3.5/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/home/users/kob/mastera/pysot/pysot/utils/distributed.py", line 43, in forward
return self.module(*args, **kwargs)
File "/home/users/kob/mastera/pysot/venv/lib/python3.5/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/home/users/kob/mastera/pysot/pysot/models/model_builder.py", line 103, in forward
loc_loss = weight_l1_loss(loc, label_loc, label_loc_weight)
File "/home/users/kob/mastera/pysot/pysot/models/loss.py", line 33, in weight_l1_loss
diff = (pred_loc - label_loc).abs()
RuntimeError: The size of tensor a (17) must match the size of tensor b (25) at non-singleton dimension 4

提个外行的问题😁尝试了一下，剪了一段视频最初ROI选定人脸，由于画面有切换，不是一直追踪拍摄，切换后目标就乱了，使用SiamMask算法，SiamRPN稍好一点。有时会再捕捉到原目标，有时就走远了~~

class MultiRPN(RPN):
    def __init__(self, anchor_num, in_channels, weighted=False):
        super(MultiRPN, self).__init__()
        self.weighted = weighted
        for i in range(len(in_channels)):
            self.add_module('rpn'+str(i+2), 
                    DepthwiseRPN(anchor_num, in_channels[i], in_channels[i]))
        if self.weighted:
            self.cls_weight = nn.Parameter(torch.ones(len(in_channels)))
            self.loc_weight = nn.Parameter(torch.ones(len(in_channels)))

    def forward(self, z_fs, x_fs):
        cls = []
        loc = []
        for idx, (z_f, x_f) in enumerate(zip(z_fs, x_fs), start=2):
            rpn = getattr(self, 'rpn'+str(idx))
            c, l = rpn(z_f, x_f)
            cls.append(c)
            loc.append(l)

        if self.weighted:
            cls_weight = F.softmax(self.cls_weight, 0)
            loc_weight = F.softmax(self.loc_weight, 0)

        def avg(lst):
            return sum(lst) / len(lst)

        def weighted_avg(lst, weight):
            s = 0
            for i in range(len(weight)):
                s += lst[i] * weight[i]
            return s

        if self.weighted:
            return weighted_avg(cls, cls_weight), weighted_avg(loc, loc_weight)
        else:
            return avg(cls), avg(loc)

where did you change the weight? It seems

cls_weight = F.softmax(self.cls_weight, 0)
loc_weight = F.softmax(self.loc_weight, 0)

will get a mean version.