feymanpriv / dolg-paddle Goto Github PK
View Code? Open in Web Editor NEWPaddle Implementation of DOLG (ICCV 2021)
Paddle Implementation of DOLG (ICCV 2021)
Hello,
Thank you for sharing the architecture of DOLG.
I have few questions regarding the implementation and the config file.
I saw in the architecture that there is no final layer used during training. Does the final descriptor is just followed by a linear layer with a softmax and arcface module ?
A question about the config Yaml, STRIDE_1X1 = False
, but in the core.config it is written : # Apply stride to 1x1 conv (True -> MSRA; False -> fb.torch)
_C.RESNET.STRIDE_1X1 = True
Does that mean, I need to change in the yaml STRIDE_1X1 to True as we are using paddle and not pytorch ?
Thanks in advance for your help!
Wonderful work!
But I got a question while reading the paper, since you didn't public the code.
In the local branch, after the Multi-Atrous module, we will get 3 feature maps with a shape of Bx512xHxW. Then the lower branch will provide a “global” feature vector with Bx512x1x1. Till now, am I right?
Then, I wonder that if the concatenation operation is conducted on the channel dim or spatial dim?
Hello @feymanpriv,
First of all, thank you for sharing good codes including this one :D
I saw the query image cropping issue on ROxford / RParis test presented by @andrefaraujo.
Using the code and R101 DOLG model weight you uploaded, I tried the ROxford5K / RParis6K experiment with query image cropping.
The results were shown as below table: (5 multi-scale inference is applied follow your paper)
ROxf-M | +1M | RPar-M | +1M | ROxf-H | +1M | RPar-H | +1M | |
---|---|---|---|---|---|---|---|---|
DOLG-paper (maybe w/o query cropping) | 81.5 | 77.4 | 91.0 | 83.3 | 61.1 | 54.8 | 80.3 | 66.7 |
DOLG-paddle-weight (w/o query cropping) | 83.5 | - | 91.8 | - | 65.3 | - | 82.8 | - |
DOLG-paddle-weight (with query cropping) | 82.2 | 74.0 | 91.1 | 80.7 | 64.4 | 51.4 | 81.9 | 63.2 |
The results of this experiment seem to indicate that the results of the paper may have been misreported.
p.s. DOLG incorporates CurricularFace Loss, not an ArcFace. However, the paper only describes ArcFace.
It seems to need correction.
Hi, if use your released r101 model, then how to preprocess the image?RGB or BGR? just divide 255 or some else ? Thanks!
如题,前几天还对着源码理解论文来着,现在咋404了...
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.