search

ShuweiShao / AF-SfMLearner

[MedIA2022 & ICRA2021] Self-Supervised Monocular Depth and Ego-Motion Estimation in Endoscopy: Appearance Flow to the Rescue
MIT License
99 stars 15 forks source link

测试集误差 #35

Closed han12580 closed 1 year ago

han12580 commented 1 year ago

您好！ 我在深度估计测试集上使用您提供的End-to-end模型进行测试，得到结果如下 image

略逊于您说的这个结果 image 请问这个误差是合理的吗？

ShuweiShao commented 1 year ago

按理说是应该一样的，不过这个误差已经很小了。
---- 回复的原邮件 ----
发件人 Juntao ***@***.***>
日期 2023年04月23日 19:02
收件人 ***@***.***>
抄送至 ***@***.***>
主题 [ShuweiShao/AF-SfMLearner] 测试集误差 (Issue #35)

您好！
我在深度估计测试集上使用您提供的End-to-end模型进行测试，得到结果如下
image

略逊于您说的这个结果
image
请问这个误差是合理的吗？

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.Message ID: <ShuweiShao/AF-SfMLearner/issues/35@github.com>

han12580 commented 1 year ago

按理说是应该一样的，不过这个误差已经很小了。





---- 回复的原邮件 ----
发件人 <a class="mail-from" style="color: #1c83eb; text-decoration: none" @." >Juntao @.>
日期 2023年04月23日 19:02
收件人 <a class="mail-to" style="color: #1c83eb; text-decoration: none" @." >@.>
抄送至 <a class="mail-cc" style="color: #1c83eb; text-decoration: none" @." >@.>
主题 [ShuweiShao/AF-SfMLearner] 测试集误差 (Issue #35)

您好！
我在深度估计测试集上使用您提供的End-to-end模型进行测试，得到结果如下
<img](https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png" alt="image" style="max-width: 100%;">

略逊于您说的这个结果
<img](https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png" alt="image" style="max-width: 100%;">
请问这个误差是合理的吗？

—
Reply to this email directly, view it on GitHub, or unsubscribe.<br](https://github.com/notifications/unsubscribe-auth/ASQFOBFR4Z7OZHWBJPWFRMLXCUD3VANCNFSM6AAAAAAXIOUD3E%22>unsubscribe.<br) />You are receiving this because you are subscribed to this thread.Message ID: span>&lt;ShuweiShao/AF-SfMLearner/issues/35</span><span>@</spangithub.com>

是的我也纳闷为什么误差那么小，我刚刚又试了试位姿估计（轨迹1），发现也是和论文的误差很小 image image

han12580 commented 1 year ago

按理说是应该一样的，不过这个误差已经很小了。





---- 回复的原邮件 ----
发件人 <a class="mail-from" style="color: #1c83eb; text-decoration: none" @." >Juntao @.>
日期 2023年04月23日 19:02
收件人 <a class="mail-to" style="color: #1c83eb; text-decoration: none" @." >@.>
抄送至 <a class="mail-cc" style="color: #1c83eb; text-decoration: none" @." >@.>
主题 [ShuweiShao/AF-SfMLearner] 测试集误差 (Issue #35)

您好！
我在深度估计测试集上使用您提供的End-to-end模型进行测试，得到结果如下
<img](https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png" alt="image" style="max-width: 100%;">

略逊于您说的这个结果
<img](https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png" alt="image" style="max-width: 100%;">
请问这个误差是合理的吗？

—
Reply to this email directly, view it on GitHub, or unsubscribe.<br](https://github.com/notifications/unsubscribe-auth/ASQFOBFR4Z7OZHWBJPWFRMLXCUD3VANCNFSM6AAAAAAXIOUD3E%22>unsubscribe.<br) />You are receiving this because you are subscribed to this thread.Message ID: span>&lt;ShuweiShao/AF-SfMLearner/issues/35</span><span>@</spangithub.com>

会是pytorch版本问题吗？我用的是1.13.0版本

ShuweiShao commented 1 year ago

有可能，之前我用的1.2.0
---- 回复的原邮件 ----
发件人 Juntao ***@***.***>
日期 2023年04月23日 20:14
收件人 ***@***.***>
抄送至 ***@***.***>、***@***.***>
主题 Re: [ShuweiShao/AF-SfMLearner] 测试集误差 (Issue #35)

按理说是应该一样的，不过这个误差已经很小了。


<div style="font-size: 16px;font-weight:bold; width:100%; white-space: nowrap; overflow:hidden;text-overflow: @.
<div style="width: 100%;font-size: 14px @.



---- 回复的原邮件 ----
发件人 <a class="mail-from" style="color: #1c83eb; text-decoration: none" @." >Juntao @.>
日期 2023年04月23日 19:02
收件人 <a class="mail-to" style="color: #1c83eb; text-decoration: none" @." >@.>
抄送至 <a class="mail-cc" style="color: #1c83eb; text-decoration: none" @." >@.>
主题 [ShuweiShao/AF-SfMLearner] 测试集误差 (Issue #35)

您好！
我在深度估计测试集上使用您提供的End-to-end模型进行测试，得到结果如下
<img](https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836065-f1696f72-0b45-4034-943b-6e67c0fe0b42.png" alt="image" style="max-width: 100%;">

略逊于您说的这个结果
<img](https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png%22><img) src="https://user-images.githubusercontent.com/94788634/233836001-8697cd41-7d7a-4431-830e-773b736e0839.png" alt="image" style="max-width: 100%;">
请问这个误差是合理的吗？

—
Reply to this email directly, view it on GitHub, or unsubscribe.<br](https://github.com/notifications/unsubscribe-auth/ASQFOBFR4Z7OZHWBJPWFRMLXCUD3VANCNFSM6AAAAAAXIOUD3E%22>unsubscribe.<br) />You are receiving this because you are subscribed to this thread.Message ID: </issues/35@github.com>

<script type="application/ld+json">[ { @.": "http://schema.org", @.": "EmailMessage", "potentialAction": { @.": "ViewAction", "target": "#35", "url": "#35", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { @.": "Organization", "name": "GitHub", "url": "https://github.com" } } ]</script>

会是pytorch版本问题吗？我用的是1.13.0版本

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: <ShuweiShao/AF-SfMLearner/issues/35/1519053233@github.com>

han12580 commented 1 year ago

好的，那应该是版本问题，非常感谢您

  • © Githubissues.
  • Githubissues is a development platform for aggregating issues.