a

app.py

@@ -1,6 +1,8 @@
 import gradio as gr
 from transformers import ViltProcessor, ViltForQuestionAnswering
 import torch
+from googletrans import Translator
+from googletrans import LANGCODES
 
 torch.hub.download_url_to_file('http://images.cocodataset.org/val2017/000000039769.jpg', 'cats.jpg')
 
@@ -23,7 +25,7 @@ def answer_question(image, text):
 image = gr.inputs.Image(type="pil")
 question = gr.inputs.Textbox(label="Question")
 answer = gr.outputs.Textbox(label="Predicted answer")
-examples = [["cats.jpg", "How many cats are there?"]]
+examples = [["cats.jpg", "How many cats are there, in French?"]]
 
 title = "Cross-lingual VQA"
 description = "ViLT (Vision and Language Transformer), fine-tuned on VQAv2 "

requirements.txt

@@ -1,3 +1,4 @@
-gradio
+gradio==2.9.1
 torch
 git+https://github.com/huggingface/transformers.git
+googletrans==4.0.0-rc1