请问能把您配置好的vot-toolkit发我一份吗？我运行一直不成功。

Hi, when run python demo.py in branch py36 :

Traceback (most recent call last):
File "/work/GIT/07_Track/SPLT/demo.py", line 782, in
tracker = MobileTracker(vot=use_vot, dis=display)
File "/work/GIT/07_Track/SPLT/demo.py", line 101, in init
= build_box_predictor(model, model_scope, self.initConstantOp, reuse=None)
File "/work/GIT/07_Track/SPLT/utils/tracking_utils.py", line 466, in build_box_predictor
prediction_dict = model.predict_box_with_init(init_feature_maps, preprocessed_images, istraining=False)
File "/work/GIT/07_Track/SPLT/core/man_meta_arch.py", line 399, in predict_box_with_init
self._anchors = self._anchor_generator.generate(feature_map_spatial_dims)
File "/work/GIT/07_Track/SPLT/lib/object_detection/core/anchor_generator.py", line 101, in generate
anchors = self._generate(feature_map_shape_list, **params)
File "/work/GIT/07_Track/SPLT/lib/object_detection/anchor_generators/multiple_grid_anchor_generator.py", line 202, in _generate
concatenated_anchors = box_list_ops.concatenate(anchor_grid_list)
File "/work/GIT/07_Track/SPLT/lib/object_detection/core/box_list_ops.py", line 529, in concatenate
raise ValueError('all elements of boxlists should be BoxList objects')
ValueError: all elements of boxlists should be BoxList objects

Hi，
Thanks for your contribution on SOT, And how long the pytorch version will be released?

 
workspace_load.m文件下第142行的文件路径换成你的VOT数据集路径。可能还需要自己写一个list.txt文件。

------------------ 原始邮件 ------------------
发件人: "feidasheng"<[email protected]>;
发送时间: 2019年11月1日(星期五) 晚上8:13
收件人: "iiau-tracker/SPLT"<[email protected]>;
抄送: "Eilo"<[email protected]>;"Author"<[email protected]>;
主题: Re: [iiau-tracker/SPLT] vot-toolkit ? (#3)

请问你解决了么，我这里测vot的数据集应该放哪呀？

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.

Originally posted by @eilo1234 in #3 (comment)

您好：
请问可以提供VOT2018 Long-trem数据集的下载地址吗？因为我在VOT官网下载不了。如果可以的话，我将十分感谢。

Hi，
Thanks for your contribution on SOT, And how long the PyTorch version will be released?
你好，请问你是否会发布PyTorch版的工程代码？

run demo have enourmous errors of arugument numbers

have never met issues like this

Use standard file APIs to check for files with this prefix.
Traceback (most recent call last):
  File "demo.py", line 782, in <module>
    tracker = MobileTracker(vot=use_vot, dis=display)
  File "demo.py", line 93, in __init__
    model = build_man_model(model_config=model_config, is_training=False)
  File "/home/jinyu/splt/SPLT-master/core/model_builder.py", line 74, in build_man_model
    hard_example_miner) = losses_builder.build(model_config.loss)
ValueError: too many values to unpack

(have added one is_training in model_builder.py)
!
PYTHON27&PYTHON36 both errororororrr

The readme don't have training skimming part.

Are you using the MIT license?

Could you please offer a link to downlad the vot-2018 long-term dataset?

When I attempted to run the SPLT in python 3.6, there was an error as "ValueError: bad marshal data (unknown type code)". I tried to solve it by deleting the .pyc file ,but it didn't work. Could you please have solution to this issue?