lzrobots / deepembeddingmodel_zsl Goto Github PK

HI, I'm currently try to implement the DeepEmbed with various CNN. I don't really understand the size of the visual feature. In the paper, it is stated that you are using 1024 dimension. But in your coding you are using 2048 which is doubled. I already check the Inception-V2. It also produce 1024 just right before the FC. Can anyone explain to me why it is 2048?

Thanks

HI, I'm currently try to implement the DeepEmbed with various CNN. I don't really understand the size of the visual feature. In the paper, it is stated that you are using 1024 dimension. But in your coding you are using 2048 which is doubled. I already check the Inception-V2. It also produce 1024 just right before the FC. Can anyone explain to me why it is 2048?

Thanks

Hi, I am trying to reproduce your results. I wait till all the iterations are completed and note the final result. However, the results do not exactly match or are very close to that reported in the home page of this Github repository. I know this can happen due to the random initialization strategy of neural networks. But, how do I get results which are sufficiently close to the reported ones. Do you use early stopping, which is not actually present in the code ? Kindly help. (I am talking about the *_GBU.py scripts)

In the tables, you compare all the methods in the conventional and generalized setting. So, is there any way to measure the statistical significance of pairs of methods ? In that case, over how many trials the averaged accuracy has been found ? If we know the accuracy results of different trials, I think we can we find the significance level (p-value). Or are they just evaluated on some fixed large batch of test dataset without any trials ?

Hi, in the dataset ImageNet2012/2010, did you use all samples in ImageNet2012 to train? And which part of ImageNet2010 was used to test, validation part or testing part or both?

Hi, I'm new in Zero_shot Learning and i'm facing a problem here

I tried to implement the AwA_Attribute.py

and I get this respond:

File "AwA_attribute.py", line 125, in
att_batch_val, visual_batch_val = iter_.next()
AttributeError: 'generator' object has no attribute 'next'