dcavadia commited on
Commit
34f59b7
Β·
1 Parent(s): 39b64b3

update info

Browse files
Files changed (1) hide show
  1. app.py +20 -14
app.py CHANGED
@@ -8,27 +8,29 @@ def create_demo():
8
 
9
  **Performance Metrics:**
10
  - **88% mAP\\@0.5** detection accuracy
11
- - **92% pixel-level** segmentation accuracy
12
  - **35+ FPS** real-time processing
13
- - **YOLOv8 + U-Net** dual architecture
14
 
15
  ### πŸ₯ Clinical Capabilities
16
- - Real-time polyp detection and localization
17
- - Precise boundary segmentation with pixel-level accuracy
18
- - Quantitative size measurement for clinical decision support
 
19
  - Optimized for endoscopy workflow integration
20
 
21
  ### 🎯 Technical Highlights
22
  - **Multi-modal AI**: Combined object detection and segmentation
23
  - **Clinical-grade performance**: Sub-second processing times
24
- - **Automated measurement**: Size, area, and morphometric analysis
 
25
  - **Real-time inference**: GPU-accelerated deployment ready
26
 
27
  ---
28
 
29
  ## πŸŽ₯ Live System Demonstration
30
 
31
- **Real-time EndoSight AI in action** - showcasing automated polyp detection, segmentation, and measurement analysis on endoscopy footage.
32
  """
33
 
34
  footer = """
@@ -49,7 +51,7 @@ def create_demo():
49
  ### πŸ“ž Professional Contact
50
 
51
  **Technical Lead**: Daniel Cavadia
52
- **Email**: dan.cavadia@gmail.com
53
 
54
  ---
55
 
@@ -64,16 +66,20 @@ def create_demo():
64
  # Simple video display (no fancy features in older version)
65
  video = gr.Video(
66
  value="demo_video.mp4",
67
- label="EndoSight AI Real-time Detection & Segmentation"
68
  )
69
 
70
  gr.Markdown("""
71
  ### πŸ” What You're Seeing:
72
- - **Blue bounding boxes**: Real-time polyp detection
73
- - **Colored masks**: Precise segmentation boundaries
74
- - **Measurement overlays**: Automated size calculations
75
- - **Processing metrics**: FPS and accuracy indicators
76
- - **Multi-polyp detection**: Simultaneous analysis capability
 
 
 
 
77
  """)
78
 
79
  gr.Markdown(footer)
 
8
 
9
  **Performance Metrics:**
10
  - **88% mAP\\@0.5** detection accuracy
11
+ - **75.6% Dice Score** segmentation accuracy
12
  - **35+ FPS** real-time processing
13
+ - **YOLOv8 + U-Net** dual architecture with intelligent stabilization
14
 
15
  ### πŸ₯ Clinical Capabilities
16
+ - Real-time polyp detection and localization with smooth tracking
17
+ - Precise boundary segmentation with fluid heatmap visualization
18
+ - Intelligent measurement system with anti-fluctuation technology
19
+ - Automated size classification and risk assessment
20
  - Optimized for endoscopy workflow integration
21
 
22
  ### 🎯 Technical Highlights
23
  - **Multi-modal AI**: Combined object detection and segmentation
24
  - **Clinical-grade performance**: Sub-second processing times
25
+ - **Intelligent measurement**: Size, area, and morphometric analysis with temporal stabilization
26
+ - **Adaptive visualization**: Real-time heatmaps that adjust to camera movement
27
  - **Real-time inference**: GPU-accelerated deployment ready
28
 
29
  ---
30
 
31
  ## πŸŽ₯ Live System Demonstration
32
 
33
+ **Real-time EndoSight AI in action** - showcasing automated polyp detection, intelligent segmentation, stabilized measurement analysis, and clinical classification on endoscopy footage.
34
  """
35
 
36
  footer = """
 
51
  ### πŸ“ž Professional Contact
52
 
53
  **Technical Lead**: Daniel Cavadia
54
+ **Email**: [dan.cavadia@gmail.com](mailto:dan.cavadia@gmail.com)
55
 
56
  ---
57
 
 
66
  # Simple video display (no fancy features in older version)
67
  video = gr.Video(
68
  value="demo_video.mp4",
69
+ label="EndoSight AI Real-time Detection & Segmentation with Intelligent Measurements"
70
  )
71
 
72
  gr.Markdown("""
73
  ### πŸ” What You're Seeing:
74
+ - **Cyan bounding boxes**: Real-time polyp detection with smooth tracking
75
+ - **Fluid heatmaps**: Dynamic segmentation visualization that adapts to movement
76
+ - **Bottom-left panels**: Stabilized measurement displays with clinical classifications
77
+ - **Size categories**: Diminutive (<5mm), Small (5-9mm), Large (β‰₯10mm) classifications
78
+ - **Risk assessment**: Low/Moderate/High risk indicators based on polyp size
79
+ - **Measurement accuracy**: Diameter and area calculations with error margins
80
+ - **Anti-fluctuation**: Intelligent system prevents measurement swings during camera movement
81
+ - **Processing metrics**: Real-time FPS and confidence indicators
82
+ - **Multi-polyp detection**: Simultaneous analysis capability with unique tracking IDs
83
  """)
84
 
85
  gr.Markdown(footer)