Update app.py

app.py

@@ -30,26 +30,31 @@ model.to(device)
 
 
 
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model.to(device)
+
+
+
 max_length = 16
 num_beams = 4
 gen_kwargs = {"max_length": max_length, "num_beams": num_beams}
-def predict_step(image_paths):
-  images = []
-  for image_path in image_paths:
-    i_image = Image.open(image_path)
+def predict_step1(image_paths):
+    i_image = PIL.Image.open(image_paths)
     if i_image.mode != "RGB":
-      i_image = i_image.convert(mode="RGB")
+        i_image = i_image.convert(mode="RGB")
 
-    images.append(i_image)
+    pixel_values = feature_extractor(images=i_image, return_tensors="pt").pixel_values
+    pixel_values = pixel_values.to(device)
 
-  pixel_values = feature_extractor(images=images, return_tensors="pt").pixel_values
-  pixel_values = pixel_values.to(device)
+    output_ids = model.generate(pixel_values, **gen_kwargs)
 
-  output_ids = model.generate(pixel_values, **gen_kwargs)
+    preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
+    preds = [pred.strip() for pred in preds]
+    return preds
+
+  
 
-  preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
-  preds = [pred.strip() for pred in preds]
-  return preds
 
 #predict_step(['/content/drive/MyDrive/caption generator/horses.png'])
 
@@ -64,20 +69,29 @@ inputs = [
     gr.inputs.Image(type="pil", label="Original Image")
 ]
 
+outputs = [
+import gradio as gr
+
+inputs = [
+    gr.inputs.Image(type="filepath", label="Original Image")
+]
+
 outputs = [
     gr.outputs.Textbox(label = 'Caption')
 ]
 
-title = "Image Captioning using ViT + GPT2"
-description = "ViT and GPT2 are used to generate Image Caption for the uploaded image. COCO Dataset was used for training. This image captioning model might have some biases that we couldn't figure during our stress testing, so if you find any bias (gender, race and so on) please use `Flag` button to flag the image with bias"
-article = " <a href='https://huggingface.co/sachin/vit2distilgpt2'>Model Repo on Hugging Face Model Hub</a>"
+title = "Image Captioning"
+description = "ViT and GPT2 are used to generate Image Caption for the uploaded image."
+article = " <a href='https://huggingface.co/nlpconnect/vit-gpt2-image-captioning'>Model Repo on Hugging Face Model Hub</a>"
 examples = [
     ["horses.png"],
-    ["persons.jpeg"],
-    ['football_player']
+    ['persons.png'],
+    ['football_player.png']
 
 ]
 
+
+
 gr.Interface(
     predict_step,
     inputs,