lombardata commited on
Commit
eb85cd6
1 Parent(s): ce4d57a

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +10 -2
app.py CHANGED
@@ -1,6 +1,5 @@
1
  import numpy as np
2
  import gradio as gr
3
- # teo
4
  import torch
5
  from transformers import Dinov2Config, Dinov2Model, Dinov2ForImageClassification, AutoImageProcessor
6
  import torch.nn as nn
@@ -27,6 +26,7 @@ class NewheadDinov2ForImageClassification(Dinov2ForImageClassification):
27
 
28
  # Classifier head
29
  self.classifier = create_head(config.hidden_size * 2, config.num_labels)
 
30
  # IMPORT CLASSIFICATION MODEL
31
  checkpoint_name = "lombardata/dino-base-2023_11_27-with_custom_head"
32
  # import labels
@@ -63,12 +63,20 @@ def predict(input_image):
63
  i += 1
64
  result = {key: result[key] for key in result if result[key] > 0.5}
65
  return result
 
 
 
 
 
 
66
 
67
  gr.Interface(
68
  fn=predict,
69
  inputs=gr.Image(shape=(224, 224)),
70
  #outputs=gr.Label(num_top_classes=5),
71
- outputs="label").launch()
 
 
72
 
73
 
74
  #examples=["GOPR0106.JPG",
 
1
  import numpy as np
2
  import gradio as gr
 
3
  import torch
4
  from transformers import Dinov2Config, Dinov2Model, Dinov2ForImageClassification, AutoImageProcessor
5
  import torch.nn as nn
 
26
 
27
  # Classifier head
28
  self.classifier = create_head(config.hidden_size * 2, config.num_labels)
29
+
30
  # IMPORT CLASSIFICATION MODEL
31
  checkpoint_name = "lombardata/dino-base-2023_11_27-with_custom_head"
32
  # import labels
 
63
  i += 1
64
  result = {key: result[key] for key in result if result[key] > 0.5}
65
  return result
66
+
67
+ # Define style
68
+ title = "DinoVd'eau image classification"
69
+ description = f"This is a prototype application that demonstrates how artificial intelligence-based systems can recognize what object(s)
70
+ is present in an underwater image. To use it, simply upload your image, or click one of the example images to load them. For predictions,
71
+ we use the open-source model {checkpoint_name}"
72
 
73
  gr.Interface(
74
  fn=predict,
75
  inputs=gr.Image(shape=(224, 224)),
76
  #outputs=gr.Label(num_top_classes=5),
77
+ outputs="label",
78
+ title=title,
79
+ description=description).launch()
80
 
81
 
82
  #examples=["GOPR0106.JPG",