测试过程中的问题

[xyl-507]

作者您好：

感谢你们的杰出贡献。最近在复现你们的代码过程中出现一个问题。我用pysot的环境（torch==0.4）来跑您们的代码，训练完成之后，只有19个checkpoint，并且在resume和test的过程中，在加载模型时都出现了assertion error。

(pysot) xyl@xyl-System-Product-Name:~/SiamCAR-master/experiments/siamcar_mobilenetv2$ python -u ../../tools/test.py --snapshot snapshot/checkpoint_e8.pth --config config.yaml --dataset UAV123
Traceback (most recent call last):
File "../../tools/test.py", line 127, in
main()
File "../../tools/test.py", line 61, in main
model = load_pretrain(model, args.snapshot).cuda().eval()
File "/home/xyl/SiamCAR-master/pysot/utils/model_load.py", line 51, in load_pretrain
map_location=lambda storage, loc: storage.cuda(device))
File "/home/xyl/anaconda3/envs/pysot/lib/python3.7/site-packages/torch/serialization.py", line 358, in load
return _load(f, map_location, pickle_module)
File "/home/xyl/anaconda3/envs/pysot/lib/python3.7/site-packages/torch/serialization.py", line 548, in _load
assert key in deserialized_objects
AssertionError

请问一下：是不是因为代码环境不一样？可是我看你们的介绍说在torch==0.4的环境中也实现了。

[xyl-507]

把torch换成1.2版本的确实可以测试了，成功运行。但现在一个问题就是，将backbone换成pysot中的mobilenetv2，训练过程没有出现问题，测试过程仍然报错：
assert key in deserialized_objects
AssertionError
请问一下，是出了什么问题吗？
@Capricorn231

[job333]

你好，可以交流一下吗，我把backbone换成alexnet，测试过程也报错了。qq1824553806
loading UAV123: 100%|██████████████████████████████████| 123/123 [00:10<00:00, 12.11it/s, person4_1]
Traceback (most recent call last):
File "test.py", line 126, in
main()
File "test.py", line 90, in main
outputs = tracker.track(img, hp)
File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\tracker\siamcar_tracker.py", line 123, in track
outputs = self.model.track(x_crop)
File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\models\model_builder.py", line 55, in track
features = self.xcorr_depthwise(xf[0],self.zf[0])
File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\utils\xcorr.py", line 44, in xcorr_depthwise
x = x.view(1, batch*channel, x.size(2), x.size(3))
IndexError: Dimension out of range (expected to be in range of [-3, 2], but got 3)

[xyl-507]

@job333 咱们的问题还不一样，你的模型已经加载成功了，我的在加载阶段就出错了。可能SiamCAR不支持更换主干网络把

[job333]

加下qq,我上面发了，再探讨一下这个问题

[job333]

@xyl-507

[WangJun-ZJUT]

你好，可以交流一下吗，我把backbone换成alexnet，测试过程也报错了。qq1824553806 loading UAV123: 100%|██████████████████████████████████| 123/123 [00:10<00:00, 12.11it/s, person4_1] Traceback (most recent call last): File "test.py", line 126, in main() File "test.py", line 90, in main outputs = tracker.track(img, hp) File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\tracker\siamcar_tracker.py", line 123, in track outputs = self.model.track(x_crop) File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\models\model_builder.py", line 55, in track features = self.xcorr_depthwise(xf[0],self.zf[0]) File "E:\siamese-rpn-pytorch\za\SiamCAR-master\pysot\utils\xcorr.py", line 44, in xcorr_depthwise x = x.view(1, batch*channel, x.size(2), x.size(3)) IndexError: Dimension out of range (expected to be in range of [-3, 2], but got 3)

您好！关于这个问题是因为在resnet中融合了多层特征，backbone返回的是多层特征的列表，更换成alexnet后返回的是tensor，所以索引上会差一个维度。

[job333]

@WangJun-ZJUT 请教您这具体该怎么改，代码能力实在是差一些，感谢

[WangJun-ZJUT]

@WangJun-ZJUT 请教您这具体该怎么改，代码能力实在是差一些，感谢

您好，在alexnet的输出套一层list就可以了

[xyl-507]

@WangJun-ZJUT 作者您好，请问一下换了mobilenetv2，训练可以，为什么测试会报错：
(pysot) xyl@xyl-System-Product-Name:~/SiamCAR-master/experiments/siamcar_mobilenetv2$ python -u ../../tools/test.py --snapshot snapshot/checkpoint_e8.pth --config config.yaml --dataset UAV123
Traceback (most recent call last):
File "../../tools/test.py", line 127, in
main()
File "../../tools/test.py", line 61, in main
model = load_pretrain(model, args.snapshot).cuda().eval()
File "/home/xyl/SiamCAR-master/pysot/utils/model_load.py", line 51, in load_pretrain
map_location=lambda storage, loc: storage.cuda(device))
File "/home/xyl/anaconda3/envs/pysot/lib/python3.7/site-packages/torch/serialization.py", line 358, in load
return _load(f, map_location, pickle_module)
File "/home/xyl/anaconda3/envs/pysot/lib/python3.7/site-packages/torch/serialization.py", line 548, in _load
assert key in deserialized_objects
AssertionError

[xyl-507]

@WangJun-ZJUT 作者您好，我用resnet50为主干网络训练的siamcar在测试阶段只有checkpoint_e10,e15,e16，可以成功测试。其它的权重就会报错：
RuntimeError: storage has wrong size: expected 5754537876681439371 got 16384
or
ValueError: unsupported pickle protocol: 128
请问一下：这种是什么原因导致的？
十分感谢！！

[WangJun-ZJUT]

抱歉，我们没有遇见过这种情况，考虑是由于硬件问题引起的。

…

---原始邮件--- 发件人: ***@***.***&gt; 发送时间: 2022年4月2日(周六) 晚上10:15 收件人: ***@***.***&gt;; 抄送: ***@***.******@***.***&gt;; 主题: Re: [ohhhyeahhh/SiamCAR] 测试过程中的问题 (Issue #77) @WangJun-ZJUT 作者您好，我用resnet50为主干网络训练的siamcar在测试阶段只有checkpoint_e10,e15,e16，可以成功测试。其它的权重就会报错： RuntimeError: storage has wrong size: expected 5754537876681439371 got 16384 or ValueError: unsupported pickle protocol: 128 请问一下：这种是什么原因导致的？ 十分感谢！！ — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***&gt;

[xyl-507]

好的，十分感谢作者，确实是你们贡献很好，忍不住想试试，无奈呀，这么晚打扰你，抱歉了！哈哈哈。 | | xueliangdongdong | | ***@***.*** 前途很远很暗，不要怕，不怕的人前面才有路。 |

…

---- 回复的原邮件 ---- | 发件人 | ***@***.***> | | 日期 | 2022年04月02日 23:35 | | 收件人 | ***@***.***> | | 抄送至 | ***@***.******@***.***> | | 主题 | Re: [ohhhyeahhh/SiamCAR] 测试过程中的问题 (Issue #77) | 抱歉，我们没有遇见过这种情况，考虑是由于硬件问题引起的。

---原始邮件--- 发件人: ***@***.***&gt; 发送时间: 2022年4月2日(周六) 晚上10:15 收件人: ***@***.***&gt;; 抄送: ***@***.******@***.***&gt;; 主题: Re: [ohhhyeahhh/SiamCAR] 测试过程中的问题 (Issue #77) @WangJun-ZJUT 作者您好，我用resnet50为主干网络训练的siamcar在测试阶段只有checkpoint_e10,e15,e16，可以成功测试。其它的权重就会报错： RuntimeError: storage has wrong size: expected 5754537876681439371 got 16384 or ValueError: unsupported pickle protocol: 128 请问一下：这种是什么原因导致的？ 十分感谢！！ — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***&gt; — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***>

[iYbei]

你好，我想问一下我在评估OTB100数据集上跑出来的结果出现了以下错误，是因为什么呢？跑出来的结果是100个txt文件
(base) at@at-RESCUER-R720-15IKBN:~/SiamCAR-master/tools$ python eval.py --tracker_path ./results --dataset OTB100 --tracker_prefix 'general_model'
Traceback (most recent call last):
File "eval.py", line 104, in
main()
File "eval.py", line 37, in main
assert len(trackers) > 0
AssertionError