Comments (5)
Details please.
- How to reproduce? What network, what version of MatConvNet?
- Full error text / back trace
from netvlad.
The problem is due to the latest version of MatConvNet (1.0-beta20) which introduced an error, thanks for pointing this out.
I have raised this issue with the MatConvNet team ( vlfeat/matconvnet#552 ) and am waiting to see what they do / how they fix it.
In the mean time, you have two options:
-
Use a slightly older version of MatConvNet (1.0-beta18 or 1.0-beta19)
-
Apply the following changes which seem to fix the problem:
a) in MatConvNet matlab/simplenn/vl_simplenn_tidy.m
change line 50 if ~isfield(layer, 'weights')
to if ~isfield(layer, 'weights') && ~isprop(layer, 'weights')
, and line 106 if ~isfield(layer, defaults{i})
to if ~isfield(layer, defaults{i}) && ~isprop(layer, defaults{i})
b) In NetVLAD code, add weights= {}
to files layerWholeL2Normalize.m
, layerTotalAvg.m
, layerTotalMax.m
just after the line precious= false
c) in NetVLAD code, file netPrepareForTrain
, change line 6 if i>=backPropToLayer && isFieldOrProp(net.layers{i}, 'weights')
to if i>=backPropToLayer && isFieldOrProp(net.layers{i}, 'weights') && ~isempty(net.layers{i}.weights)
from netvlad.
Sorry for I did not give the details.
-
setting:
NetVLAD: Version 1.03 (04 Mar 2016)
relja_matlab: Version 1.02 (29 Feb 2016)
MatConvNet: 1.0-beta20 -
Error:
There are some errors when I use the function "relja_simplenn_tidy".Ther information:
'''''''''''''''''''''''''''''''''''
No public field weights exists for class layerWholeL2Normalize.
Error in vl_simplenn_tidy (line 51)
layer.weights = {};
Error in relja_simplenn_tidy (line 25)
net = vl_simplenn_tidy(net);
Error in test (line 21)
net = relja_simplenn_tidy(net);
'''''''''''''''''''''''''''''''''''''''''''
3.Test Code:
% Set the MATLAB paths
setup;
% ---------- Use/test our networks
% Load our network
netID= 'vd16_tokyoTM_conv5_3_vlad_preL2_intra_white';
paths= localPaths();
load( sprintf('%s%s.mat', paths.ourCNNs, netID), 'net' );
a = sprintf('%s%s.mat', paths.ourCNNs, netID);
disp (a);
net= relja_simplenn_tidy(net); //error
4.And I am new with MATLAB.. Thank you very much!!
from netvlad.
thank you very much!
from netvlad.
The MatConvNet team has fixed the issue I referenced above (in patch vlfeat/matconvnet@91399d4 ), so the NetVLAD code works without any modifications now.
from netvlad.
Related Issues (20)
- the TokyoTM database HOT 2
- pitts30k datasets HOT 9
- pitts30k datasets HOT 1
- Dataset of NetVLAD HOT 1
- Applying to Indoor usage HOT 2
- about Pittsburgh 250k dataset HOT 1
- Link to download Pittsburgh 250k: available on request here showing invalid link HOT 11
- Converting MATLAB trained model .mat files (off the shelf and trained on Pittsburg 30k) in Pytorch HOT 1
- Pitts30k datasets HOT 1
- NetVLAD effectiveness in Gazebo Environments HOT 3
- Questions about Pitts250k dataset panoramic image splitting HOT 1
- How to run the Trained Network Matlab HOT 1
- using in real time application HOT 1
- About demo test HOT 4
- some problem of dem.m HOT 14
- What is the difference between these two NetVLAD programs? HOT 4
- Where can I get Pitts250k data? HOT 1
- Pitts30k Datasets! HOT 1
- About datasets HOT 1
- Is it possible to convert to ONNX format? HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from netvlad.