pulkitag / ief Goto Github PK
View Code? Open in Web Editor NEWIterative Error Feedback
Iterative Error Feedback
I tried the demo and got the error below.
[libprotobuf ERROR google/protobuf/text_format.cc:245] Error parsing text-format caffe.NetParameter: 133:28: Message type "caffe.PoolingParameter" has no field named "round_below".
WARNING: Logging before InitGoogleLogging() is written to STDERR
F0708 03:12:45.061872 17362 upgrade_proto.cpp:932] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: models/vgg_s.prototxt
*** Check failure stack trace: ***
This issue is similar to #3947 .
I use the latest caffe.
My environment is below.
Ubuntu 14.04+CUDA7.0+cudnn4.0
Thank you in advance.
The demo only works for 1 example/batch at the moment. Upgrade it to work with batchsize > 1.
in class pseIEF, self.mxStepSz_ = metaPose['mxStepNorm'] , it is used to multiply with the correction distance. What does it do? How to get it in the training phase? Thanks very much.
I tried demo.py
.
I got the error below.
I0708 12:26:11.132272 22880 net.cpp:413] Input 0 -> image
I0708 12:26:11.132319 22880 net.cpp:413] Input 1 -> kp_pos
I0708 12:26:11.132338 22880 net.cpp:413] Input 2 -> label
I0708 12:26:11.132359 22880 layer_factory.hpp:77] Creating layer render1
F0708 12:26:11.132725 22880 layer_factory.hpp:81] Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: Python (known types: AbsVal, Accuracy, ArgMax, BNLL, BatchNorm, BatchReindex, Bias, Concat, ContrastiveLoss, Convolution, CropData, CroppingKat, Data, Deconvolution, Dropout, DummyData, ELU, Eltwise, Embed, EuclideanLoss, EuclideanLossWithIgnore, Exp, Filter, Flatten, GenericWindowData, GenericWindowData2, GlobalEltwise, HDF5Data, HDF5Output, HingeLoss, Im2col, ImageData, InfogainLoss, InnerProduct, LRN, Log, MVN, MemoryData, MultinomialLogisticLoss, Normalize, PReLU, Pooling, Power, ROIPooling, RandomNoise, ReLU, Reduction, Reshape, SPP, Scale, Sigmoid, SigmoidCrossEntropyLoss, Silence, Slice, SmoothL1Loss, Softmax, SoftmaxWithLoss, Split, SquareBoxData, TanH, Threshold, Tile, Topography, WindowData)
*** Check failure stack trace: ***
I'm using the pulkitag/caffe.
I got no error with make all
.
But I got the error below with make test
.
CXX src/caffe/test/test_data_layer.cpp
/usr/bin/g++ src/caffe/test/test_data_layer.cpp -MMD -MP -pthread -fPIC -DCAFFE_VERSION=1.0.0-rc3 -DNDEBUG -O2 -DUSE_OPENCV -DUSE_LEVELDB -DUSE_LMDB -I/usr/include/python2.7 -I/usr/lib/python2.7/dist-packages/numpy/core/include -I/usr/local/include -I.build_release/src -I./src -I./include -I/usr/local/cuda-7.0/include -Wall -Wno-sign-compare -c -o .build_release/src/caffe/test/test_data_layer.o 2> .build_release/src/caffe/test/test_data_layer.o.warnings.txt
|| (cat .build_release/src/caffe/test/test_data_layer.o.warnings.txt; exit 1)
src/caffe/test/test_roi_pooling_layer.cpp:19:38: fatal error: caffe/fast_rcnn_layers.hpp: No such file or directory
#include "caffe/fast_rcnn_layers.hpp"
^
compilation terminated.
make: *** [.build_release/src/caffe/test/test_roi_pooling_layer.o] ERROR 1
My environment is below.
ubuntu 14.04+CUDA7.0+cudnn4.0
Could you help me?
Thank you in advance.
When I execute the following
ief = td.PoseIEF()
in a ipython notebook, I am getting the following error.
File "/home/drive/Softwares/pulkitag/ief/src/pycaffe_utils/my_pycaffe.py", line 286, in setup_network
self.net = caffe.Net(self.defFile_, self.modelFile_, caffe.TEST)
SystemError: NULL result without error in PyObject_Call
I was trying to run the demo. I am getting the following error:
WARNING: Logging before InitGoogleLogging() is written to STDERR
F0808 09:21:26.621508 2574 common.cpp:53] CPU-only Mode: cannot make GPU call.
*** Check failure stack trace: ***
Aborted (core dumped)
I have even set isGPU= False
Hi, I want to use the IEF to train with my own data, how can do it?
I try to input the data like this:
input: "image"
input_dim: 1
input_dim: 3
input_dim: 224
input_dim: 224
input: "kp_pos"
input_dim: 1
input_dim: 17
input_dim: 2
input_dim: 1
input: "label"
input_dim: 1
input_dim: 16
input_dim: 2
input_dim: 1
but I got the error as:
I0221 09:49:03.289826 27112 caffe.cpp:219] Starting Optimization
I0221 09:49:03.289835 27112 solver.cpp:280] Solving
I0221 09:49:03.289839 27112 solver.cpp:281] Learning Rate Policy: poly
Using Config:
Namespace(K=100.0, T=-50.0, imgSz=224, sigma=0.001)
Using Config:
Namespace(K=100.0, T=-50.0, imgSz=224, sigma=0.001)
Traceback (most recent call last):
File "/path../caffe_ief/python_layer/src/layers/python_ief.py", line 71, in forward
top[0].data[b][k,yImSt:yImEn,xImSt:xImEn] = copy.deepcopy(self.g_[ySt:yEn, xSt:xEn])
ValueError: could not broadcast input array from shape (410,224) into shape (0,224)
Hello,
It would be grateful, if you provide us with training codes,
Thanks
Based on my understanding, if for instance the output is 1D, the ground truth y is 5 and the initial y0 is 1. We will use 4 steps to regress the data. So we will regress the fixed corrections [4, 3, 2, 1] (epsilon:5-1, 5-2, 5-3, 5-4) for every iteration. For every step, we will update the current out concated with the input image. In the end, the output should be y0 + epsilon0 + epsilon1+epsilon2+epsilon3. If so, the target will be not reasonable because 4+3+2+1 != 5-1. Otherwise, is the fixed corrections selected according to y0+y1+y2+y3 = 4? How to set the fixed corrections? Thanks very much.
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.