Oversampling

[Seamus Mackenzie]

I want to run a test on a pre-trained model with the validation dataset from ILSVRC2014


for example in pycaffe :-

net = caffe.Net("/home/seamus/Downloads/BVLC_GoogLeNet/train_val.prototxt",
                "/home/seamus/Downloads/BVLC_GoogLeNet/snapshot_iter_67.caffemodel",
                caffe.TEST)


accuracy = 0
IterationTest=10

for i in range(IterationTest):
  # one iteration
  outputs = net.forward()
  accuracy+=net.blobs['accuracy-top5'].data
  print outputs

print('========================================================================')
avgAcc=accuracy/IterationTest
print 'accuracy = ', avgAcc

I want to do exactly this but with oversampling as in caffe.classifyer.predict(oversample=True) I want this to validate Top-5 Accuracy of the network on the whole dataset but using oversampling.

Is there any built in way to do this or can someone with greater knowledge than me who has done this before provide a script?

I've been searching for hours but can only find instances where it was used to classify batches of images rather than measuring accuracy


Thanks

Seamus