qianyiwu / objsdf Goto Github PK

Hi~
Congrats on your excellent job! I wanna ask whether you have added some procesing on nerf++. In your code, I saw the word, inverse_sphere_bg. But i didn't find the related network for background.

Could u release this part of job?
Thanks!

Hi Qianyi,

Thanks for your wonderful work.

I have some questions about the semantic field s in your paper. Why should the partial derivative ∂s/∂d meet the desired requirement (smooth inside but a rapid change at the boundary) rather than the function s? Thanks!

Hi,
我想请教一下如何生成ObjectNeRF场景编辑时，所需的bbox.json文件，或者能解释一下bbox.json里面各个数值的含义吗？
下面是bbox.json的内容
"labels": [ { "id": "1", "data": { "position": [ 2.2837844942793737, 0.6488003671313685, 6.001685665511459 ], "rotation": [ 1.8946632971215467, -0.1095624676165341, 0.553559251247026 ], "quaternion": [ 0.771054293372373, -0.2522670140662573, 0.11654286129474664, 0.5729348931312866 ], "scale": [ 3.9938567122761546, 1.853964693824193, 1.8692240780858735 ] } },

Hi!
Congratulations on your wonderful work!
I meet some problem with the experiment on scannet dataset. I see the code needs "label_mapping_instance.txt", and I get it from "scenexxxx_00.aggregation.json" as "0, 1, 2, .....,9 ", but there are still errors when running code, as "nll_loss_forward_reduce_cuda_kernel_2d: block: [0,0,0], thread: [28,0,0] Assertion t >= 0 && t < n_classes failed."
It seems that the semantic (ground truth) has something wrong.
I wonder what is wrong with it.
Thank you in advance!

Toydesk链接出现错误，可以重新分享下数据集么

Thanks for the outstanding work!
I found that the training time is very long, It needs 120 hours for 10000 epochs training on 2080ti.
Can you provide the checkpoints, It would help me a lot.

My best wish.

Hi, thank you for sharing your wonderful work.
I've noticed that you've finished your test on ScanNet dataset, could you provide that data, code, and checkpoints?

Hi Qianyi, I wonder why you use "SDF bbox" to clamp SDFs. Could you explain this? I haven't find any description in paper.

It seems like we can convert SDFs of different objects to semantic fields via sigmoid.

Hi, Thanks for your amazing work! I'm curious about the content in your center.txt. From your code, the content is related to scale_mat and pose_matrix. As my own understanding it can help generate the sample sphere, could you please describe it clearly? Thank you for your reply, best wishes!

Hi Qianyi,

The link of Toydesk is out of date. May I have new sharing link?

Hi Qianyi,

I didn't find the corresponding script of mesh evaluation on Toydesk dataset in your codebase, and also in the objnerf codebase. I wonder how did you get the Chamfer Distance provided in the paper?

Because in the toydesk dataset, it seems that the gt point cloud is with many outliers, I wonder if there is any other operations that need to be done in addition to aligning the output mesh and gt with center and scale.

Hi Qian Wu,

Thank you for your help in dataset and Sorry for that I still have problems in evaluation process.

• I have trained a model with toydesk2, it takes 15 hours in training process (2000 epoch with 3080Ti), and I wonder how long it takes you to train the model for toydesk?
• I find the ply files in the exp folder (default output folder), but their quality is not that good. I want to know how I can obtain the mesh on the quality that you displayed. Following is the color mesh I generate based on a training output in final epoch.
  
• when I execute the eval file, I find there are many attributions that the ToydeskDataset don't have such as get_scale_mat() function and scan_id. Moreover, I wonder what scan_id means? I defined it as -1, and it doesn't report any error but give me a 'killed' print.

Thank you for your taking time to going through my questions and wait for your reply.
Best Regards