adding self_caption

app.py

@@ -42,26 +42,28 @@ model = VisionEncoderDecoderModel.from_pretrained("./vit-bert")
 
 '''
 
-repo_name = "ydshieh/vit-gpt2-coco-en"
-#test_image = "cats.jpg"
-url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
-test_image = Image.open(requests.get(url, stream=True).raw)
-test_image.save("cats.png")
-feature_extractor2 = ViTFeatureExtractor.from_pretrained(repo_name)
-tokenizer = AutoTokenizer.from_pretrained(repo_name)
-model2 = VisionEncoderDecoderModel.from_pretrained(repo_name)
-pixel_values = feature_extractor2(test_image, return_tensors="pt").pixel_values
-print("Pixel Values")
-print(pixel_values)
-# autoregressively generate text (using beam search or other decoding strategy)
-generated_ids = model2.generate(pixel_values, max_length=16, num_beams=4, return_dict_in_generate=True)
-# decode into text
-preds = tokenizer.batch_decode(generated_ids[0], skip_special_tokens=True)
-preds = [pred.strip() for pred in preds]
-print("Predictions")
-print(preds)
 
 
+def self_caption(image):
+  repo_name = "ydshieh/vit-gpt2-coco-en"
+  #test_image = "cats.jpg"
+  url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
+  test_image = Image.open(requests.get(url, stream=True).raw)
+  test_image.save("cats.png")
+  feature_extractor2 = ViTFeatureExtractor.from_pretrained(repo_name)
+  tokenizer = AutoTokenizer.from_pretrained(repo_name)
+  model2 = VisionEncoderDecoderModel.from_pretrained(repo_name)
+  pixel_values = feature_extractor2(test_image, return_tensors="pt").pixel_values
+  print("Pixel Values")
+  print(pixel_values)
+  # autoregressively generate text (using beam search or other decoding strategy)
+  generated_ids = model2.generate(pixel_values, max_length=16, num_beams=4, return_dict_in_generate=True)
+  # decode into text
+  preds = tokenizer.batch_decode(generated_ids[0], skip_special_tokens=True)
+  preds = [pred.strip() for pred in preds]
+  print("Predictions")
+  print(preds)
+  return preds
 
 def classify_image(image):
   results = image_pipe(image)
@@ -87,8 +89,8 @@ title = "Generate a Story from an Image"
 description = "Demo for classifying images with Perceiver IO. To use it, simply upload an image and click 'submit' to let the model predict the 5 most probable ImageNet classes. Results will show up in a few seconds."  + image_piped
 article = "<p style='text-align: center'></p>"
 
-gr.Interface(fn=classify_image, inputs=image, outputs=label, title=title, description=description, examples="", enable_queue=True).launch(debug=True)
-
+#gr.Interface(fn=classify_image, inputs=image, outputs=label, title=title, description=description, examples="", enable_queue=True).launch(debug=True)
+gr.Interface([classify_image,self_caption], inputs=image, outputs=label, title=title, description=description, examples="", enable_queue=True).launch(debug=True)