Update app.py

app.py

@@ -1,5 +1,8 @@
 import gradio as gr
 import streamlit as st
+import torch 
+import re 
+from transformers import AutoTokenizer, ViTFeatureExtractor, VisionEncoderDecoderModel 
 
 # def greet(name):
 #     return "Hello " + name + "!!"
@@ -7,6 +10,23 @@ import streamlit as st
 # iface = gr.Interface(fn=greet, inputs="text", outputs="text")
 # iface.launch()
 
+device='cpu'
+encoder_checkpoint = "nlpconnect/vit-gpt2-image-captioning"
+decoder_checkpoint = "nlpconnect/vit-gpt2-image-captioning"
+model_checkpoint = "nlpconnect/vit-gpt2-image-captioning"
+feature_extractor = ViTFeatureExtractor.from_pretrained(encoder_checkpoint)
+tokenizer = AutoTokenizer.from_pretrained(decoder_checkpoint)
+model = VisionEncoderDecoderModel.from_pretrained(model_checkpoint).to(device)
+
+
+def predict(image,max_length=64, num_beams=4):
+  image = image.convert('RGB')
+  image = feature_extractor(image, return_tensors="pt").pixel_values.to(device)
+  clean_text = lambda x: x.replace('<|endoftext|>','').split('\n')[0]
+  caption_ids = model.generate(image, max_length = max_length)[0]
+  caption_text = clean_text(tokenizer.decode(caption_ids))
+  return caption_text 
+
 st.title("Image to Text using Lora")
 
 inputs = gr.inputs.Image(label="Upload any Image", type = 'pil', optional=True)
@@ -15,7 +35,7 @@ description = "NTT Data Bilbao team"
 title = "Image to Text using Lora"
 
 interface = gr.Interface(
-        # fn=predict,
+        fn=predict,
         description=description,
         inputs = inputs,
         theme="grass",