|
| 1 | +# Metadata |
| 2 | +author: Y-T-G |
| 3 | +task: segment |
| 4 | +keywords: [cnn, yolo11, small-object] |
| 5 | +description: YOLO11 instance segmentation model with an extra P2 scale for small object detecton. |
| 6 | +flops: |
| 7 | + n: 25.4 |
| 8 | + s: 80.5 |
| 9 | + m: 284.6 |
| 10 | + l: 309.2 |
| 11 | + x: 683.9 |
| 12 | +parameters: |
| 13 | + n: 2993984 |
| 14 | + s: 10235568 |
| 15 | + m: 22194640 |
| 16 | + l: 27719888 |
| 17 | + x: 61446576 |
| 18 | +min_version: 8.3.0 |
| 19 | + |
| 20 | +# Model |
| 21 | +nc: 80 |
| 22 | +strides: [4, 8, 16, 32] |
| 23 | +scales: |
| 24 | + n: [0.50, 0.25, 1024] |
| 25 | + s: [0.50, 0.50, 1024] |
| 26 | + m: [0.50, 1.00, 512] |
| 27 | + l: [1.00, 1.00, 512] |
| 28 | + x: [1.00, 1.50, 512] |
| 29 | + |
| 30 | +backbone: |
| 31 | + # Stem |
| 32 | + - [-1, 1, Conv, [64, 3, 2]] # (1, 64, 320, 320) - 0 |
| 33 | + |
| 34 | + # P2 Backbone |
| 35 | + - [-1, 1, Conv, [128, 3, 2]] # (1, 128, 160, 160) - 1 |
| 36 | + - [-1, 2, C3k2, [256, False, 0.25]] # (1, 256, 160, 160) - 2 |
| 37 | + |
| 38 | + # P3 Backbone |
| 39 | + - [-1, 1, Conv, [256, 3, 2]] # (1, 256, 80, 80) - 3 |
| 40 | + - [-1, 2, C3k2, [512, False, 0.25]] # (1, 512, 80, 80) - 4 |
| 41 | + |
| 42 | + # P4 Backbone |
| 43 | + - [-1, 1, Conv, [512, 3, 2]] # (1, 512, 40, 40) - 5 |
| 44 | + - [-1, 2, C3k2, [512, True]] # (1, 512, 40, 40) - 6 |
| 45 | + |
| 46 | + # P5 Backbone |
| 47 | + - [-1, 1, Conv, [1024, 3, 2]] # (1, 512, 20, 20) - 7 |
| 48 | + - [-1, 2, C3k2, [1024, True]] # (1, 512, 20, 20) - 8 |
| 49 | + - [-1, 1, SPPF, [1024, 5]] # (1, 512, 20, 20) - 9 |
| 50 | + - [-1, 2, C2PSA, [1024]] # (1, 512, 20, 20) - 10 |
| 51 | + |
| 52 | +head: |
| 53 | + # Top-Down Path (Feature Pyramid) |
| 54 | + # P4 Head |
| 55 | + - [-1, 1, nn.Upsample, [None, 2, nearest]] # (1, 512, 40, 40) - 11 |
| 56 | + - [[-1, 6], 1, Concat, [1]] # (1, 1024, 40, 40) - 12 |
| 57 | + - [-1, 2, C3k2, [512, False]] # (1, 512, 40, 40) - 13 |
| 58 | + |
| 59 | + # P3 Head |
| 60 | + - [-1, 1, nn.Upsample, [None, 2, nearest]] # (1, 512, 80, 80) - 14 |
| 61 | + - [[-1, 4], 1, Concat, [1]] # (1, 1024, 80, 80) - 15 |
| 62 | + - [-1, 2, C3k2, [256, False]] # (1, 256, 80, 80) - 16 |
| 63 | + |
| 64 | + # P2 Head |
| 65 | + - [-1, 1, nn.Upsample, [None, 2, nearest]] # (1, 256, 160, 160) - 17 |
| 66 | + - [[-1, 2], 1, Concat, [1]] # (1, 512, 160, 160) - 18 |
| 67 | + - [-1, 2, C3k2, [128, False]] # (1, 128, 160, 160) - 19 |
| 68 | + |
| 69 | + # Bottom-Up Path (PAN-FPN) |
| 70 | + # P3 |
| 71 | + - [-1, 1, Conv, [128, 3, 2]] # (1, 128, 80, 80) - 20 |
| 72 | + - [[-1, 16], 1, Concat, [1]] # (1, 384, 80, 80) - 21 |
| 73 | + - [-1, 2, C3k2, [256, False]] # (1, 256, 80, 80) - 22 |
| 74 | + |
| 75 | + # P4 |
| 76 | + - [-1, 1, Conv, [256, 3, 2]] # (1, 256, 40, 40) - 23 |
| 77 | + - [[-1, 13], 1, Concat, [1]] # (1, 768, 40, 40) - 24 |
| 78 | + - [-1, 2, C3k2, [512, False]] # (1, 512, 40, 40) - 25 |
| 79 | + |
| 80 | + # P5 |
| 81 | + - [-1, 1, Conv, [512, 3, 2]] # (1, 512, 20, 20) - 26 |
| 82 | + - [[-1, 10], 1, Concat, [1]] # (1, 1024, 20, 20) - 27 |
| 83 | + - [-1, 2, C3k2, [1024, True]] # (1, 512, 20, 20) - 28 |
| 84 | + |
| 85 | + # Segmentation Head (P2, P3, P4, P5) |
| 86 | + - [[19, 22, 25, 28], 1, Segment, [nc, 32, 256]] # 29 |
0 commit comments