Pretrained-Show-and-Tell-model icon indicating copy to clipboard operation
Pretrained-Show-and-Tell-model copied to clipboard

Model does not produce expected results

Open vineetrshenoy opened this issue 5 years ago • 5 comments

I have followed the directions in the repository to download and run the model. The code successfully executes for the following image (directly from the Tensorflow im2txt repository) COCO_val2014_000000224477

However, when I run the model against this image, I receive the intended results:

Screenshot from 2019-05-03 10-47-55

What could be the problem that I am running into?

vineetrshenoy avatar May 03 '19 14:05 vineetrshenoy

I have the same issue. Also, I get the same out of memory warnings you're getting on my 1080, and I had to add allow_growth config option or it would bomb.

I also tried it on a rented 2080 ti and got the same classification, but no memory warnings.

ianmcmahon avatar May 05 '19 18:05 ianmcmahon

Same probleme here, any update yet? Also tried this implementation here: https://modeldepot.io/leeryboy/show-and-tell but the results are the same.

coliinkc avatar Jun 08 '19 17:06 coliinkc

Hello, I have the same issue of poor performance. Can you tell me what is the problem

xieyxclack avatar Jun 10 '19 05:06 xieyxclack

Found the solution:

https://github.com/tensorflow/models/issues/6513

coliinkc avatar Jun 11 '19 17:06 coliinkc

When I change line 182 to: most_likely_words = np.argsort(word_probabilities)[-self.beam_size][::-1]

I get: IndexError: invalid index to scalar variable.

Did anyone face this or know how to solve it?

Edit

I changed line 182 to: most_likely_words = np.argsort(word_probabilities)[::-1] and that worked for me.

senorpinatta avatar Jul 16 '19 05:07 senorpinatta