always returns the same caption

[ada-dl]

I used the pretrained checkpoint and wordmap, when I ran it through different images, it always returns the same caption? Why could this be? Could you please help me?

[maotaitas]

I had the same problem and commenting out img = imresize(img, (256, 256)) [line 36 in caption.py], thus not resizing the input resolved the issue.

[fragno]

It's better to add img_as_ubyte to change img array from float to integer.

img = img_as_ubyte(imresize(img, (256, 256)))

[nsharan2000]

that

It's better to add img_as_ubyte to change img array from float to integer.

img = img_as_ubyte(imresize(img, (256, 256)))

this helped perfectly, thank you