How should I visualize the test results?

[ljsandlbc]

I was at 《test.sh》 adds -- debug 4 , but an error is reported. The content is [cv2. error: OpenCV (4.6.0)/io/opencv/modules/core/src/algorithm. cpp: 212: error: (- 209: Sizes of input arguments do not match) The operation is next 'array op array' (where arrays have the same size and type), nor 'array op scalar', nor 'scalar op array' in function 'binary_op']

[ChristophLanglitz]

Hi, have you found a solution? I have the same problem

[ljsandlbc]

我已经解决这个问题了，是在代码进行了简单修改，但是这个代码是我去年改的，我得找一下，后续有情况了我告诉你

…

------------------ 原始邮件 ------------------ 发件人: "brandesjj/centerfusionpp" ***@***.***>; 发送时间: 2023年6月1日(星期四) 晚上9:33 ***@***.***>; ***@***.******@***.***>; 主题: Re: [brandesjj/centerfusionpp] How should I visualize the test results? (Issue #2) Hi, have you found a solution? I have the same problem — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[ChristophLanglitz]

Thank you! Would help me a lot.

I have something else. When I use the original CenterFusion repository by mrnabati and run the <test.sh> and use the pretrained models (--load_model) then I get great detection results with --debug 1. But if I train a model by myself with the <train.sh> and I try to --load_model it into the <test.sh> it just shows me the images without any bounding boxes. Do you know why? Do I train it wrong?

[ljsandlbc]

我明白了你的问题，貌似我是没有这个问题的，因为我记得我当时测试的时候因为训练轮数较少，最后的检测框虽然不准确，但确实是有框的。如果你想讨论更多的内容，可以直接给我邮箱发信息，因为在别人问题下面水贴而没有实际回复感觉怪不好意思的，hhh。

…

---Original--- From: ***@***.***&gt; Date: Thu, Jun 1, 2023 22:19 PM To: ***@***.***&gt;; Cc: ***@***.******@***.***&gt;; Subject: Re: [brandesjj/centerfusionpp] How should I visualize the testresults? (Issue #2) Thank you! :) Would help me a lot. I have something else. When I use the original CenterFusion repository by mrnabati and run the <test.sh&gt; and use the pretrained models (--load_model) then I get great detection results with --debug 1. But if I train a model by myself with the <train.sh&gt; and I try to --load_model it into the <test.sh&gt; it just shows me the images without any bounding boxes. Do you know why? Do I train it wrong? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***&gt;

[ChristophLanglitz]

I have opened a new issue. If you want you can answer there.
Would be much appreciated!

[Manishnayak234]

@ChristophLanglitz Hey can you tell me the what are the versions of lib you have downloaded, can you provide docker images or list of lib version for the whole repo.