jimgoo / caffe-oxford102 Goto Github PK
View Code? Open in Web Editor NEWCaffe CNNs for the Oxford 102 flower dataset
Caffe CNNs for the Oxford 102 flower dataset
Hi,
Thanks for sharing the experiments info, I was wondering that if could give me a link to pretrained-weight for VGG_S finetune part?
the layer is:
layers {
name: "data"
type: IMAGE_DATA
top: "data"
top: "label"
image_data_param {
#source: "/home/ubuntu/git/caffe-oxford102/train.txt"
source: "/home/ubuntu/git/caffe-oxford102/test.txt" # Flipped
batch_size: 50
new_height: 256
new_width: 256
}
transform_param {
crop_size: 227
mean_file: "/home/ubuntu/caffe/data/ilsvrc12/imagenet_mean.binaryproto"
mirror: true
}
include: { phase: TRAIN }
}
erro is:
Unknown layer type: ImageData (known types: AbsVal, Accuracy, ArgMax, BNLL, BatchNorm, BatchReindex, Bias, Concat, ContrastiveLoss, Convolution, Crop, Data, Deconvolution, Dropout, DummyData, ELU, Eltwise, Embed, EuclideanLoss, Exp, Filter, Flatten, HDF5Data, HDF5Output, HingeLoss, Im2col, InfogainLoss, InnerProduct, Input, LRN, LSTM, LSTMUnit, Log, MVN, MemoryData, MultinomialLogisticLoss, PReLU, Parameter, Pooling, Power, Python, RNN, ReLU, Reduction, Reshape, SPP, Scale, Sigmoid, SigmoidCrossEntropyLoss, Silence, Slice, Softmax, SoftmaxWithLoss, Split, TanH, Threshold, Tile)
Thanks a lot for great work!
Please could you let me know what mean numbers and channel order I should use to pre-process images?
I fine tuned the AlexNet with the training set, and the loss divergent, but when using the test set(6k images) instead, it converged. So I'm really curious about how you fine tuned the AlexNet and got a good result. : )
Hi,
may I ask that how the splits.png was generated?
Thanks!
Thanks for the nice tutorial! However, the AlexNet model cannot be downloaded -- https://s3.amazonaws.com/jgoode/cannaid/bvlc_reference_caffenet.caffemodel returns "forbidden" or "access denied".
holle, please tell me the difference between test_acc and test_acc_flipped?? thank you!!
I am trying to implement the same project but using Tensorflow, and I was surprised when you were able to achieve 80% accuracy after only 500 iterations. May I ask how did you normalize the images before training? Did you divide them by 255 or subtract the ImageNet mean?
Thank you
Hi.
Since http://zeus.robots.ox.ac.uk/flower_demo/demo currently doesn't work, I am trying your model.
I wrote the following script, but it outputs almost same percentage (about 1%) probabilities of all labels even if the input is the one in the data set.
Could you tell me my misunderstanding to use your model?
Thanks,
import numpy as np
import caffe
classifier = caffe.Classifier("<git folder>/caffe-oxford102/AlexNet/deploy.prototxt", "oxford102.caffemodel")
image = caffe.io.load_image("<git folder>/caffe-oxford102/data/jpg/image_00348.jpg")
inputs = [image]
predictions = classifier.predict(inputs, oversample=False)
print(predictions)
I use the same way as you and obtained a good performance,but when I use a single picture to test the model it only obtained 0.0098 accuracy.....please tell me why?
Hi, How can I get my own dataset format, like the imagelabels.mat and setid.mat if I want train my own models, Thx
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.