huanglianghua / globaltrack Goto Github PK

Thank you for the author's code. Do we need to preprocess the coco, LaSOT, GOT10K data when reproducing the code ourselves, or just put it in the corresponding folder?

Hi,
Could you please provide your result files(LaSOT, TLP..) available?

have you plan to release your code??

Traceback (most recent call last):
File "tools/test_global_track.py", line 1, in
import _init_paths
ModuleNotFoundError: No module named '_init_paths'

I followed the steps for the setup. How to fix this?

The evaluation results of GlobalTrack on DTB70 seem not satisfying.
I've changed the output format before evaluation. Any suggestions for such results?

I saw the library neuron in this repo GlobalTrack, it seems to an advanced version of the repo got10k for visual tracking dataset processing. So, as the title says, do you have a plan to release that repo named
as neuron? thank you and all of your doing.

as title, thanks

following the tutorial in readme.md, python setup.py develop
But there is no mmcv in the conda list.
What is the corresponding mmcv version required? Can I install it from ?
Thank you!

I found this model:“faster_rcnn_r50_fpn_2x_20181010-443129e1.pth” from mmdetection(v1.0rc1)(https://github.com/open-mmlab/mmdetection/blob/v1.0rc1/docs/MODEL_ZOO.md) . Did the initial weights you provided(qg_rcnn_r50_fpn_2x_20181010-443129e1.pth) transferred from this model? And how to do that?

Thank you!

hi,
I got some error:
ImportError: _submodules/mmdetection/mmdet/ops/dcn/deform_conv_cuda.cpython-37m-x86_64-linux-gnu.so: undefined symbol: __cudaPopCallConfiguration
I tried on two computer(1060 and 2080ti), they all got same error.
Can anyone help me?
thanks

你好：
我在1060, 2080ti上 嘗試運作，都獲得同樣的error，能否給予我一點協助，非常感謝

`Traceback (most recent call last):

File "tools/test_global_track.py", line 6, in
import _init_paths
File "/home/huchenjie/CODE/GlobalTrack-master/_init_paths.py", line 6, in
from modules import *
File "/home/huchenjie/CODE/GlobalTrack-master/modules/init.py", line 1, in
from .modulators import *
File "/home/huchenjie/CODE/GlobalTrack-master/modules/modulators.py", line 3, in
from mmdet.models.roi_extractors import SingleRoIExtractor
File "_submodules/mmdetection/mmdet/models/init.py", line 1, in
from .anchor_heads import * # noqa: F401,F403
File "_submodules/mmdetection/mmdet/models/anchor_heads/init.py", line 1, in
from .anchor_head import AnchorHead
File "_submodules/mmdetection/mmdet/models/anchor_heads/anchor_head.py", line 8, in
from mmdet.core import (AnchorGenerator, anchor_target, delta2bbox, force_fp32,
File "_submodules/mmdetection/mmdet/core/init.py", line 6, in
from .post_processing import * # noqa: F401, F403
File "_submodules/mmdetection/mmdet/core/post_processing/init.py", line 1, in
from .bbox_nms import multiclass_nms
File "_submodules/mmdetection/mmdet/core/post_processing/bbox_nms.py", line 3, in
from mmdet.ops.nms import nms_wrapper
File "_submodules/mmdetection/mmdet/ops/init.py", line 2, in
from .dcn import (DeformConv, DeformConvPack, DeformRoIPooling,
File "_submodules/mmdetection/mmdet/ops/dcn/init.py", line 1, in
from .deform_conv import (DeformConv, DeformConvPack, ModulatedDeformConv,
File "_submodules/mmdetection/mmdet/ops/dcn/deform_conv.py", line 9, in
from . import deform_conv_cuda
ImportError: libcudart.so.9.0: cannot open shared object file: No such file or directory
`
Thank you for the code.I installed the environment as required and successfully compiled the Cpp/CUDA extensions.The only difference is that pytorch 1.1.0 was replaced by pytorch 1.5.1.The above error occurred while I was running 'test_global_track.py'，I tried many ways but failed to solve the problem，so I hope to get help from the author and other friends.

I follow Non-distributed training to train all dataset. but when some time, the programe will interrupt. The question is IndexError: list index out of range. Can you tell me how to solve this question? Is the dataset is something wrong? @huanglianghua Thank you very much

如下图所示，请问这是为啥

_submodules/neuron/neuron/data/datasets/vot.py中的__init__函数，download设置为了False，运行时依旧会直接下载数据集

hello,thanks for your nice work!
however, i am thinking could i set more than one init_boxes on the first frame?
In another word, may this job work on MOT,can you give some advice about that?
Thanks!

mmdet和mmcv的版本问题，尝试了多个版本，一直不可以

运行程序test_global_track.py出现以下问题：
File "C:\ProgramData\Anaconda3\envs\GlobalTrack\lib\ctypes_init_.py", line 367, in init
self._handle = _dlopen(self._name, mode)
OSError: [WinError 126] 找不到指定的模块。

当我执行setup进行编译的时候，会报错：error: unknown file type '' (from 'src/soft_nms_cpu.pyx/mmdet/ops/nms')
请问这是什么情况呀，麻烦告诉我一下，谢谢您

Thanks for your excellent work and your sharing. When I download your pretrained weights(qg_rcnn_r50_fpn_coco_got10k_lasot.pth) and testing it directly on OTB100, I found the acc is only 60.6, is that something wrong? Have you ever test it on OTB100. Waiting for your reply. Best wishes!

您好，我已经改成了这个，Cuda compilation tools, release 9.0, V9.0.176
请问为什么还是会报错，有什么解决办法吗

hi,
I have two issues when I run python test_global_track.py

1. the testing data basketball, bird... in OTB-15 not performing well.
   I used original set in test_global_track.py.
   cfg_file = 'configs/qg_rcnn_r50_fpn.py'
ckp_file = 'checkpoints/qg_rcnn_r50_fpn_coco_got10k_lasot.pth'
   The problem of tracking the wrong object is serious (eg., wrong player, wrong bird... ).

2. I got the error OSError: /home/myAccount/data/LaSOTBenchmark/airplane/airplane-1/groundtruth.txt not found

Where to download the LaSOTB benchmark？
or it will download automatically, when I run test_global_track.py.

Files already downloaded.
Processing sequence [1/280]: airplane-1...
Traceback (most recent call last):
File "test_global_track.py", line 15, in
data.EvaluatorLaSOT(frame_stride=10),
File "_submodules/neuron/neuron/data/evaluators/otb_eval.py", line 430, in init
dataset = datasets.LaSOT(root_dir, subset='test')
File "_submodules/neuron/neuron/data/datasets/lasot.py", line 42, in init
subset=self.subset)
File "_submodules/neuron/neuron/data/datasets/dataset.py", line 30, in init
seq_dict = self._construct_seq_dict(**kwargs)
File "_submodules/neuron/neuron/data/datasets/lasot.py", line 69, in _construct_seq_dict
anno_files[s], delimiter=',', dtype=np.float32)
File "/home/myAccount/miniconda3/envs/globalTrack/lib/python3.7/site-packages/numpy/lib/npyio.py", line 962, in loadtxt
fh = np.lib._datasource.open(fname, 'rt', encoding=encoding)
File "/home/myAccount/miniconda3/envs/globalTrack/lib/python3.7/site-packages/numpy/lib/_datasource.py", line 266, in open
return ds.open(path, mode, encoding=encoding, newline=newline)
File "/home/myAccount/miniconda3/envs/globalTrack/lib/python3.7/site-packages/numpy/lib/_datasource.py", line 624, in open
raise IOError("%s not found." % path)
OSError: /home/myAccount/data/LaSOTBenchmark/airplane/airplane-1/groundtruth.txt not found.

Result of attributes -- 'GlobalTrack'
'ALL' overlap : 10.5% failures : 10.0
'BC' overlap : 10.0% failures : 10.0
'DEF' overlap : 9.9% failures : 10.0
'FM' overlap : 11.2% failures : 10.0
'IPR' overlap : 11.0% failures : 9.9
'IV' overlap : 10.8% failures : 10.0
'LR' overlap : 9.0% failures : 10.0
'MB' overlap : 10.9% failures : 10.0
'OCC' overlap : 9.5% failures : 10.0
'OPR' overlap : 10.2% failures : 10.0
'OV' overlap : 10.9% failures : 10.0
'SV' overlap : 10.8% failures : 10.0

mmcv版本到底用哪个呀

6fps is too slow,and it's runing on a powerful TitanX GPU.

Thank you for sharing！
Do you need additional conditions to compile successfully？such as the version of CUDA and VS .
(I chose the second option)

assuming dataset storied in ./data
What do these data sets consist of? Training set, test set?
Can you explain the directory structure?
Thank you!

I met cuda out of memory when using 2080 with 11G memory.Is there any way to solve this problem?

As the title describes, can you release your tracking results on common datasets?
Though the paper shows overall results on most long-term datasets, I want to analyze the result on each video of common datasets, especially short-term datasets, such as OTB2015, VOT2018, .etc.
Thank you!
Best regards!

Hi, thanks for your cool work!

For the RPN_Modulator class in modulators.py file, it seems that it's different with the equation (1) in the paper in 2 aspects:

• fx(x) is missing, no 3x3 conv applied to the search img feature.
• convolution between fx(x) and fz(z) is replaced by dot product.

The related code in modulators.py file is
out_ij = [self.proj_modulator[k](query) * gallary[k] for k in range(len(gallary))]

Could you check, thanks!

My GPU is 2080Ti ,CUDA version is 9.0 ,pytorch is 1.1.0, torchvisio is 0.3.0,but I meet a error:,I tried many ways but failed to solve the problem，so I hope to get help from the author and other friends.

Args:
-- Namespace(autoscale_lr=True, base_dataset='got10k_train', base_transforms='extra_partial', config='configs/qg_rcnn_r50_fpn.py', fp16=False, gpus=2, launcher='none', load_from=None, local_rank=0, resume_from=None, sampling_prob='0.4,0.4,0.2', seed=None, validate=False, work_dir='work_dirs/qg_rcnn_r50_fpn', workers=None)
Configs:
-- Config (path: /media/hdc/data4/wxl/GlobalTrack/configs/qg_rcnn_r50_fpn.py): {'model': {'type': 'QG_RCNN', 'pretrained': 'torchvision://resnet50', 'backbone': {'type': 'ResNet', 'depth': 50, 'num_stages': 4, 'out_indices': (0, 1, 2, 3), 'frozen_stages': 1, 'style': 'pytorch'}, 'neck': {'type': 'FPN', 'in_channels': [256, 512, 1024, 2048], 'out_channels': 256, 'num_outs': 5}, 'rpn_head': {'type': 'RPNHead', 'in_channels': 256, 'feat_channels': 256, 'anchor_scales': [8], 'anchor_ratios': [0.5, 1.0, 2.0], 'anchor_strides': [4, 8, 16, 32, 64], 'target_means': [0.0, 0.0, 0.0, 0.0], 'target_stds': [1.0, 1.0, 1.0, 1.0], 'loss_cls': {'type': 'CrossEntropyLoss', 'use_sigmoid': True, 'loss_weight': 1.0}, 'loss_bbox': {'type': 'SmoothL1Loss', 'beta': 0.1111111111111111, 'loss_weight': 1.0}}, 'bbox_roi_extractor': {'type': 'SingleRoIExtractor', 'roi_layer': {'type': 'RoIAlign', 'out_size': 7, 'sample_num': 2}, 'out_channels': 256, 'featmap_strides': [4, 8, 16, 32]}, 'bbox_head': {'type': 'SharedFCBBoxHead', 'num_fcs': 2, 'in_channels': 256, 'fc_out_channels': 1024, 'roi_feat_size': 7, 'num_classes': 2, 'target_means': [0.0, 0.0, 0.0, 0.0], 'target_stds': [0.1, 0.1, 0.2, 0.2], 'reg_class_agnostic': False, 'loss_cls': {'type': 'CrossEntropyLoss', 'use_sigmoid': False, 'loss_weight': 1.0}, 'loss_bbox': {'type': 'SmoothL1Loss', 'beta': 1.0, 'loss_weight': 1.0}}}, 'train_cfg': {'rpn': {'assigner': {'type': 'MaxIoUAssigner', 'pos_iou_thr': 0.7, 'neg_iou_thr': 0.3, 'min_pos_iou': 0.3, 'ignore_iof_thr': -1}, 'sampler': {'type': 'RandomSampler', 'num': 256, 'pos_fraction': 0.5, 'neg_pos_ub': -1, 'add_gt_as_proposals': False}, 'allowed_border': 0, 'pos_weight': -1, 'debug': False}, 'rpn_proposal': {'nms_across_levels': False, 'nms_pre': 2000, 'nms_post': 2000, 'max_num': 2000, 'nms_thr': 0.7, 'min_bbox_size': 0}, 'rcnn': {'assigner': {'type': 'MaxIoUAssigner', 'pos_iou_thr': 0.5, 'neg_iou_thr': 0.5, 'min_pos_iou': 0.5, 'ignore_iof_thr': -1}, 'sampler': {'type': 'RandomSampler', 'num': 512, 'pos_fraction': 0.25, 'neg_pos_ub': -1, 'add_gt_as_proposals': True}, 'pos_weight': -1, 'debug': False}}, 'test_cfg': {'rpn': {'nms_across_levels': False, 'nms_pre': 1000, 'nms_post': 1000, 'max_num': 1000, 'nms_thr': 0.7, 'min_bbox_size': 0}, 'rcnn': {'score_thr': 0.0, 'nms': {'type': 'nms', 'iou_thr': 0.5}, 'max_per_img': 1000}}, 'data': {'imgs_per_gpu': 1, 'workers_per_gpu': 4, 'train': {'type': 'PairWrapper', 'ann_file': None, 'base_dataset': 'got10k_train', 'base_transforms': 'extra_partial', 'sampling_prob': [0.4, 0.4, 0.2], 'max_size': 30000, 'max_instances': 8, 'with_label': True}}, 'optimizer': {'type': 'SGD', 'lr': 0.0025, 'momentum': 0.9, 'weight_decay': 0.0001}, 'optimizer_config': {'grad_clip': {'max_norm': 35, 'norm_type': 2}}, 'lr_config': {'policy': 'step', 'warmup': 'linear', 'warmup_iters': 500, 'warmup_ratio': 0.3333333333333333, 'step': [8, 11]}, 'checkpoint_config': {'interval': 1}, 'log_config': {'interval': 50, 'hooks': [{'type': 'TextLoggerHook'}]}, 'total_epochs': 12, 'cudnn_benchmark': True, 'dist_params': {'backend': 'nccl'}, 'log_level': 'INFO', 'work_dir': 'work_dirs/qg_rcnn_r50_fpn', 'load_from': 'checkpoints/qg_rcnn_r50_fpn_2x_20181010-443129e1.pth', 'resume_from': None, 'workflow': [('train', 1)], 'gpus': 2}
2021-02-10 12:04:44,722 - INFO - Distributed training: False
2021-02-10 12:04:45,171 - INFO - load model from: torchvision://resnet50
2021-02-10 12:04:45,366 - WARNING - The model and loaded state dict do not match exactly

unexpected key in source state_dict: fc.weight, fc.bias

2021-02-10 12:04:50,101 - INFO - load checkpoint from checkpoints/qg_rcnn_r50_fpn_2x_20181010-443129e1.pth
2021-02-10 12:04:50,431 - INFO - Start running, host: root@hdc-IBM, work_dir: /media/hdc/data4/wxl/GlobalTrack/work_dirs/qg_rcnn_r50_fpn
2021-02-10 12:04:50,432 - INFO - workflow: [('train', 1)], max: 12 epochs
THCudaCheck FAIL file=/pytorch/aten/src/THC/THCGeneral.cpp line=383 error=11 : invalid argument
Traceback (most recent call last):
File "/media/hdc/data4/wxl/GlobalTrack/tools/train_qg_rcnn.py", line 143, in
main()
File "/media/hdc/data4/wxl/GlobalTrack/tools/train_qg_rcnn.py", line 138, in main
logger=logger)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/apis/train.py", line 62, in train_detector
_non_dist_train(model, dataset, cfg, validate=validate)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/apis/train.py", line 229, in _non_dist_train
runner.run(data_loaders, cfg.workflow, cfg.total_epochs)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/mmcv/runner/runner.py", line 358, in run
epoch_runner(data_loaders[i], **kwargs)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/mmcv/runner/runner.py", line 264, in train
self.model, data_batch, train_mode=True, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/apis/train.py", line 38, in batch_processor
losses = model(**data)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 152, in forward
outputs = self.parallel_apply(replicas, inputs, kwargs)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 162, in parallel_apply
return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)])
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 83, in parallel_apply
raise output
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 59, in _worker
output = module(*input, **kwargs)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/core/fp16/decorators.py", line 49, in new_func
return old_func(*args, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/modules/qg_rcnn.py", line 58, in forward
img_z, img_x, img_meta_z, img_meta_x, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/modules/qg_rcnn.py", line 91, in forward_train
for x_ij, i, j in self.rpn_modulator(z, x, gt_bboxes_z):
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/modules/modulators.py", line 37, in forward
modulator=self.learn(feats_z, gt_bboxes_z))
File "/media/hdc/data4/wxl/GlobalTrack/modules/modulators.py", line 54, in learn
feats_z[:self.roi_extractor.num_inputs], rois)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/core/fp16/decorators.py", line 127, in new_func
return old_func(*args, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/submodules/mmdetection/mmdet/models/roi_extractors/single_level.py", line 105, in forward
roi_feats_t = self.roi_layers[i](feats[i], rois)
File "/home/hdc/anaconda3/envs/GlobalTrack_private/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/ops/roi_align/roi_align.py", line 80, in forward
self.sample_num)
File "/media/hdc/data4/wxl/GlobalTrack/_submodules/mmdetection/mmdet/ops/roi_align/roi_align.py", line 26, in forward
sample_num, output)
RuntimeError: cuda runtime error (30) : unknown error at mmdet/ops/roi_align/src/roi_align_kernel.cu:145

I'm trying to reproduce the code. I‘m just getting into PyTorch, I'd lile to ask if I need to download the dataset used in the paper? If so, how can I get the dataset? Thanks!