How to get a story for random images about glacnet HOT 4 CLOSED

Of course. However, some code modification is required.
First, follow the instructions in the README.
After that you need to modify the inference.ipynb file so that it can read the images you want.
Please note that we plan to upload additional code for this in the near future.

from glacnet.

We have tried to run the inference.ipynb file using google colab. Here's a link to the notebook: https://colab.research.google.com/drive/1P9QICcVD15aTXgOQOsa1xktH2C_ddigq?usp=sharing

I have mounted my google drive to the notebook, and a folder named "sample_test" in the drive contains the images that I want to use for testing. Each time I run the code, the output text seems to change but the images remain blacked out. All the necessary files - encoder, decoder, vocab pickle files and the json files are present in the drive.

In the screenshot below you can see the paths set for image_dir and other variables:

The Output:

The images in the sample_test folder are:

from glacnet.

I added the 'test.ipynb' file for your case.
But I couldn't test it because I don't have your photo files.
Try generating text using the 'test.ipynb' file.
Please refer to the changed code below.

image_files = ['929665.jpg', '929669.jpg', '929673.jpg', '929678.jpg', '929689.jpg']
for i in range(5):
image = Image.open('sample_test/' + image_files[i]).convert('RGB')
images.append(image)
image = transform_image(image, transform)
image_tensor.append(image)

from glacnet.

We implemented test.ipynb and we could generate stories from random images. Thank you so much for helping us out.

from glacnet.