Add gitignore

.gitignore

@@ -0,0 +1,3 @@
+checkpoints/
+env_vqa/
+

app.py

@@ -27,8 +27,8 @@ use_cuda = torch.cuda.is_available()
 # use fp16 only when GPU is available
 use_fp16 = False
 
-os.system('wget https://huggingface.co/komleva/VQAmodel/resolve/main/checkpoint.best_score_0.9840.pt; '
-          'mkdir -p checkpoints; mv checkpoint.best_score_0.9840.pt checkpoints/checkpoint.best_score_0.9840.pt')
+#os.system('wget https://huggingface.co/komleva/VQAmodel/resolve/main/checkpoint.best_score_0.9840.pt; '
+#          'mkdir -p checkpoints; mv checkpoint.best_score_0.9840.pt checkpoints/checkpoint.best_score_0.9840.pt')
 
 # specify some options for evaluation
 parser = options.get_generation_parser()
@@ -169,7 +169,7 @@ def open_domain_vqa(Image, Question):
             left = left -10
         
 
-    return Image.crop((left, top, right, bottom))#result[0]['answer']
+    return Image.crop((left, top, right, bottom)), result[0]['answer']
 
 
 title = "Visual Question Answering"
@@ -177,7 +177,7 @@ description = " Demo for Visual Question Answering. That model achieved 3rd plac
 #article = "<p style='text-align: center'><a href='https://github.com/OFA-Sys/OFA' target='_blank'>OFA Github " \
 #          "Repo</a></p> "
 examples = [['demo_imgs/000000001093.jpg', 'Which is different from the group?'], ['demo_imgs/000000001964.jpg','What do we drive for personal use?'], ['demo_imgs/000000002473.jpg','What do we use to slide on snow?']]
-io = gr.Interface(fn=open_domain_vqa, inputs=[gr.inputs.Image(type='pil'), "textbox"], outputs=gr.outputs.Image(type='pil'),#gr.outputs.Textbox(label="Answer"),
+io = gr.Interface(fn=open_domain_vqa, inputs=[gr.inputs.Image(type='pil'), "textbox"], outputs=[gr.outputs.Image(type='pil'),gr.outputs.Textbox(label="Answer")],
                   title=title, description=description, examples=examples,
                   allow_flagging=False, allow_screenshot=False)
 io.launch()