mizhenxing / gbi-net Goto Github PK
View Code? Open in Web Editor NEWCodes for GBi-Net (CVPR2022)
Home Page: https://mizhenxing.github.io/gbinet
License: MIT License
Codes for GBi-Net (CVPR2022)
Home Page: https://mizhenxing.github.io/gbinet
License: MIT License
First of all, thank you very much for your selfless sharing, this is an outstanding work.
Could you please tell me the difference between "forward_one_depth" and "forward_all_depth" in detail, and if I want to use " forward_all_depth", what do I need to change in the code, in other words: is it possible to do it in the config file?
I'd appreciate it if you could reply!
Thanks for your great work. As for the depth generation, you choose argmax of classification instead of weighed operation of regression. Does it increase the risk to predict a wrong depth range in the early stage? What do you think about the choice of classification and regression in multi-view stereo.
您好
您的论文中提到对DTU数据集使用了两种评价指标,第二种是直接评估深度图预测的准确性。
请问这部分是您自己提出的一种评价方法,还是类似第一种基于距离的评价指标,由其他方法(如DTU)提供的评估方式。
谢谢!
您好!您是否对比过coarse-to-fine search与Binary search时间方面的消耗呢?如果Binary search较快,会快出多少呢?期待您的解答。
Sorry to bother again.
In recent days, I am trying to train your network with gradient accumulation. However, my implementation still doesn‘t work,i.e., the training loss does not decrease.
I would be very appreciate if you could help provide the code about training with gradient accumulation.
Thanks and look forward to your reply!
Thanks for your significiant work. I would like to ask the method of upsampling the depth hypotheses. Is it the nearest interpolation or the bilinear interpolation? I think both of them would influence the accuarcy, especially in edge regions.
Look forward for your reply!
Hi,
I try to download the dataset several times, but I always get rejected with info. below:
Sorry, you can't view or download this file at this time.
Too many users have viewed or downloaded this file recently. Please try accessing the file again later. If the file you are trying to access is particularly large or is shared with many people, it may take up to 24 hours to be able to view or download the file. If you still can't access a file after 24 hours, contact your domain administrator.
Any solution to fix that ? Thanks!
Congratulations, what a excellent job!
But I have a question about sample interval.
For dtu_io.py line 526-527 in datasets folder:
sample_interval = depth_range / 4.0
sample_depth = []
for i in range(4):
sample_depth.append(np.ones_like(depth) * (sample_interval * (i + i + 1) / 2.0 + depth_min))
The sample depth acutally is a decimal, Why can not be an integer like this:
sample_interval = depth_range / 4.0
sample_depth = []
for i in range(4):
sample_depth.append(np.ones_like(depth) * (sample_interval * ( i+ 1) + depth_min))
Look forward your reply!
Sincerely!
Hi, I want to fuse the predicted depth maps into a point cloud(*.ply) using Gipuma fusion.
Your code based on xy-filter,,, Could you tell me how to use gipuma fusion in your project?
非常nice的工作,dtu上提升很大,并且大幅减少了训练显存,迫不及待将您的工作作为baseline了,请问什么时候可以开源代码呢?
Thanks for releasing the code!
I would like to ask about the reason of using Rectified_raw in dtu_yao?
As you coded in dtu yao, the original img is first read from the "Rectified_raw" folder, and then croped. This process can be regarded as the preprocessing of MVSNet, which delivers us a "Rectified" folder. So any difference between the above two process.
And can I directly use the "Rectified" folder from MVSNet. (Forgive me my laziness that I don't want to download the 123GB Rectified_raw hhh)
Look forward for your reply!
thanks a lot
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.