search

ThibaultGROUEIX / AtlasNet

This repository contains the source codes for the paper "AtlasNet: A Papier-Mâché Approach to Learning 3D Surface Generation ". The network is able to synthesize a mesh (point cloud + connectivity) from a low-resolution point cloud, or from an image.
http://imagine.enpc.fr/~groueixt/atlasnet/
MIT License
678 stars 120 forks source link

poor performance in my own trained model #52

Closed dangchengliang closed 4 years ago

dangchengliang commented 4 years ago

Hi,Thank you for your great works and i benefit so much from it when I read your paper .when I run your code ,and trained the datasets of shapenet,I got the same fscore as you but got poor models compared to yours.I didn't change your parameter or anyohther codes. Can you tell me what's wrong witn my operate.I sincerely hope you can give me some help. My Best wishes!

zyz-notebooks commented 4 years ago

What is your pytorch version, my 1.6 does not work

dangchengliang commented 4 years ago

My pytorch version is 1.6. The project works well in this environment.

------------------ 原始邮件 ------------------ 发件人: "ThibaultGROUEIX/AtlasNet" <notifications@github.com>; 发送时间: 2020年9月6日(星期天) 下午5:06 收件人: "ThibaultGROUEIX/AtlasNet"<AtlasNet@noreply.github.com>; 抄送: "熙熙"<2897913140@qq.com>;"Author"<author@noreply.github.com>; 主题: Re: [ThibaultGROUEIX/AtlasNet] poor performance in my own trained model (#52)

What is your pytorch version, my 1.6 does not work

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

ThibaultGROUEIX commented 4 years ago

Hi @dangchengliang , It's hard to say. Could you share a qualitative example? Did you train the single-view example and the autoencoder? Best wishes, Thibault

dangchengliang commented 4 years ago

Hi @dangchengliang , It's hard to say. Could you share a qualitative example? Did you train the single-view example and the autoencoder? Best wishes, Thibault

dangchengliang commented 4 years ago

 I have trained the autoencoder using the command 'python train.py --shapenet13',and didn't change any things of your code.I got a fscore of 0.819.in addition, I didn't train the single-view example.I attach a model of AtlasnetReconstruction to the attachment. I got this ply file through the command 'python train.py --demo' and I changed the path of 'network.pth' of my own.   I think I may have done the wrong step.  Looking forward to your reply.Thanks.

------------------ 原始邮件 ------------------ 发件人: "ThibaultGROUEIX/AtlasNet" <notifications@github.com>; 发送时间: 2020年9月6日(星期天) 晚上7:30 收件人: "ThibaultGROUEIX/AtlasNet"<AtlasNet@noreply.github.com>; 抄送: "熙熙"<2897913140@qq.com>;"Mention"<mention@noreply.github.com>; 主题: Re: [ThibaultGROUEIX/AtlasNet] poor performance in my own trained model (#52)

Hi @dangchengliang , It's hard to say. Could you share a qualitative example? Did you train the single-view example and the autoencoder? Best wishes, Thibault

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.