Update app.py

app.py

@@ -1,19 +1,18 @@
 import gradio as gr
 import torch
-from pathlib import Path
 from PIL import Image
 import pandas as pd
 from lavis.models import load_model_and_preprocess
 from lavis.processors import load_processor
-from transformers import CLIPTokenizerFast, CLIPModel  # Import CLIPTokenizerFast
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM  # Import AutoTokenizer and AutoModelForSeq2SeqLM
 
 # Load model and preprocessors for Image-Text Matching (LAVIS)
 device = torch.device("cuda") if torch.cuda.is_available() else "cpu"
 model_itm, vis_processors, text_processors = load_model_and_preprocess("blip2_image_text_matching", "pretrain", device=device, is_eval=True)
 
-# Load model and processor for Image Captioning (TextCaps)
-model_caption = CLIPModel.from_pretrained("microsoft/git-large-r-textcaps")
-processor_caption = CLIPTokenizerFast.from_pretrained("microsoft/git-large-r-textcaps", from_slow=True)  # Convert tokenizer
+# Load model and tokenizer for Image Captioning (TextCaps)
+model_caption = AutoModelForSeq2SeqLM.from_pretrained("microsoft/git-large-r-textcaps")
+tokenizer_caption = AutoTokenizer.from_pretrained("microsoft/git-large-r-textcaps")
 
 # List of statements for Image-Text Matching
 statements = [
@@ -46,9 +45,9 @@ def compute_itm_scores(image):
 # Function to generate image captions using TextCaps
 def generate_image_captions(image):
     pil_image = Image.fromarray(image.astype('uint8'), 'RGB')
-    inputs = processor_caption(pil_image, return_tensors="pt", padding=True, truncation=True)
+    inputs = tokenizer_caption(pil_image, return_tensors="pt", padding=True, truncation=True)
     outputs = model_caption.generate(**inputs)
-    caption = processor_caption.decode(outputs[0])
+    caption = tokenizer_caption.decode(outputs[0])
     return caption
 
 # Main function to perform image captioning and image-text matching
@@ -63,8 +62,9 @@ def process_images_and_statements(image):
     output = "Image Captions:\n" + captions + "\n\nITM Scores:\n" + itm_scores
     return output
 
+# Gradio interface
 image_input = gr.inputs.Image()
 output = gr.outputs.Textbox(label="Results")
 
 iface = gr.Interface(fn=process_images_and_statements, inputs=image_input, outputs=output, title="Image Captioning and Image-Text Matching")
-iface.launch()
+iface.launch()