✂️ Foreground Segmentation

Semantic segmentation for pet foreground extraction

0.92

Best IoU

4

Models

96.8%

Pixel Accuracy

U-Net

Best Model

🎯 Explore All Models

Compare different semantic segmentation architectures and approaches

🏆

U-Net

Encoder-decoder architecture with skip connections for precise segmentation

IoU
0.92
Pixel Acc
96.8%
Training Time
3.2h
Model Size
31MB
✓ Best ✓ Skip Connections
🔬

DeepLabV3+

Atrous convolution with ASPP for multi-scale feature extraction

IoU
0.89
Pixel Acc
95.2%
Training Time
4.1h
Model Size
58MB
✓ ASPP ✓ Multi-scale
🔍

PSPNet

Pyramid Scene Parsing Network with global context aggregation

IoU
0.87
Pixel Acc
94.5%
Training Time
3.8h
Model Size
65MB
✓ Global Context ✓ Pyramid
⚡

FCN

Fully Convolutional Network - End-to-end pixel-wise prediction

IoU
0.84
Pixel Acc
93.1%
Training Time
2.5h
Model Size
25MB
✓ Lightweight ✓ FCN