krishnapal2308 commited on
Commit
35e87cb
·
verified ·
1 Parent(s): 4e40bc0

Basic model's label and description changed.

Browse files
Files changed (1) hide show
  1. app.py +5 -5
app.py CHANGED
@@ -17,7 +17,7 @@ This project aims to develop a deep learning model to verbally describe image co
17
  # Define solution overview
18
  solution_overview = """
19
  ### Solution Overview
20
- The basic model, trained for a limited duration without extensive hyperparameter tuning, primarily focuses on exploring the integration of the attention mechanism with the Encoder-Decoder architecture for image processing. To improve inference quality, Vit-GPT2 architecture is integrated. [Visit the Kaggle notebook](https://www.kaggle.com/code/krishna2308/eye-for-blind) for implementation details.
21
  """
22
 
23
  # Define real-life scenario application
@@ -30,7 +30,7 @@ While this current implementation may not support real-time processing, the pote
30
  def process_image_and_generate_output(image, model_selection):
31
  if image is None:
32
  return "Please select an image", None
33
- if model_selection == 'Basic Model':
34
  result = inference_script.evaluate(image)
35
  pred_caption = ' '.join(result).rsplit(' ', 1)[0]
36
  pred_caption = pred_caption.replace('<unk>', '')
@@ -57,16 +57,16 @@ def process_image_and_generate_output(image, model_selection):
57
 
58
  sample_images = [
59
  [os.path.join(os.path.dirname(__file__), "sample_images/1.jpg"), "ViT-GPT2"],
60
- [os.path.join(os.path.dirname(__file__), "sample_images/1.jpg"), 'Basic Model'],
61
  [os.path.join(os.path.dirname(__file__), "sample_images/3.jpg"), "ViT-GPT2"],
62
- [os.path.join(os.path.dirname(__file__), "sample_images/3.jpg"), 'Basic Model']
63
  ]
64
 
65
  # Create a dropdown to select sample image
66
  image_input = gr.Image(label="Upload Image")
67
 
68
  # Create a dropdown to choose the model
69
- model_selection_input = gr.Radio(["Basic Model",
70
  "ViT-GPT2"],
71
  label="Choose Model")
72
 
 
17
  # Define solution overview
18
  solution_overview = """
19
  ### Solution Overview
20
+ The basic model, **trained for only 20 epochs without extensive hyperparameter tuning,** primarily focuses on exploring the integration of the attention mechanism with the Encoder-Decoder architecture for image processing utilizing subclassing. To improve inference quality, Vit-GPT2 architecture is integrated. [Visit the Kaggle notebook](https://www.kaggle.com/code/krishna2308/eye-for-blind) for implementation details.
21
  """
22
 
23
  # Define real-life scenario application
 
30
  def process_image_and_generate_output(image, model_selection):
31
  if image is None:
32
  return "Please select an image", None
33
+ if model_selection == 'Basic Model (Results won't be good)':
34
  result = inference_script.evaluate(image)
35
  pred_caption = ' '.join(result).rsplit(' ', 1)[0]
36
  pred_caption = pred_caption.replace('<unk>', '')
 
57
 
58
  sample_images = [
59
  [os.path.join(os.path.dirname(__file__), "sample_images/1.jpg"), "ViT-GPT2"],
60
+ [os.path.join(os.path.dirname(__file__), "sample_images/1.jpg"), 'Basic Model (Results won't be good)'],
61
  [os.path.join(os.path.dirname(__file__), "sample_images/3.jpg"), "ViT-GPT2"],
62
+ [os.path.join(os.path.dirname(__file__), "sample_images/3.jpg"), 'Basic Model (Results won't be good)']
63
  ]
64
 
65
  # Create a dropdown to select sample image
66
  image_input = gr.Image(label="Upload Image")
67
 
68
  # Create a dropdown to choose the model
69
+ model_selection_input = gr.Radio(["Basic Model (Results won't be good)",
70
  "ViT-GPT2"],
71
  label="Choose Model")
72