Implement CBAM and Involution block; Create the yaml file

harry-sunhao · Nov 22, 2023 · e3b3932 · e3b3932
1 parent 7dc89d8
commit e3b3932
Show file tree

Hide file tree

Showing 60 changed files with 1,475 additions and 63 deletions.
diff --git a/prediction_visual.png b/prediction_visual.png
diff --git a/run_notebook.bat b/run_notebook.bat
@@ -0,0 +1,9 @@
+@echo off
+REM 初始化Anaconda环境
+CALL D:\Programming\Anaconda3\Scripts\activate.bat
+REM 激活conda环境
+CALL activate pytorch
+REM 启动Jupyter Notebook
+CALL jupyter notebook --NotebookApp.token='f425885e3f44deefa51613af98fea11694b2806e9628b070'
+REM 保持窗口打开
+pause
diff --git a/run_tensorboard.bat b/run_tensorboard.bat
@@ -0,0 +1,9 @@
+@echo off
+REM 初始化Anaconda环境
+CALL D:\Programming\Anaconda3\Scripts\activate.bat
+REM 激活conda环境
+CALL activate pytorch
+REM 启动Jupyter Notebook
+CALL python -m tensorboard.main --logdir='small_grabage/'
+REM 保持窗口打开
+pause
diff --git a/small_grabage/CBAM-P2/args.yaml b/small_grabage/CBAM-P2/args.yaml
@@ -0,0 +1,98 @@
+task: detect
+mode: train
+model: yolov8n-CBAM-P2.yaml
+data: data.yaml
+epochs: 1000
+patience: 50
+batch: 16
+imgsz: 256
+save: true
+save_period: -1
+cache: false
+device: null
+workers: 8
+project: small_grabage
+name: CBAM-P2
+exist_ok: false
+pretrained: true
+optimizer: auto
+verbose: true
+seed: 0
+deterministic: true
+single_cls: false
+rect: false
+cos_lr: false
+close_mosaic: 10
+resume: null
+amp: true
+fraction: 1.0
+profile: false
+freeze: null
+overlap_mask: true
+mask_ratio: 4
+dropout: 0.0
+val: true
+split: val
+save_json: false
+save_hybrid: false
+conf: null
+iou: 0.7
+max_det: 300
+half: false
+dnn: false
+plots: true
+source: null
+show: false
+save_txt: false
+save_conf: false
+save_crop: false
+show_labels: true
+show_conf: true
+vid_stride: 1
+stream_buffer: false
+line_width: null
+visualize: false
+augment: false
+agnostic_nms: false
+classes: null
+retina_masks: false
+boxes: true
+format: torchscript
+keras: false
+optimize: false
+int8: false
+dynamic: false
+simplify: false
+opset: null
+workspace: 4
+nms: false
+lr0: 0.01
+lrf: 0.01
+momentum: 0.937
+weight_decay: 0.0005
+warmup_epochs: 3.0
+warmup_momentum: 0.8
+warmup_bias_lr: 0.1
+box: 7.5
+cls: 0.5
+dfl: 1.5
+pose: 12.0
+kobj: 1.0
+label_smoothing: 0.0
+nbs: 64
+hsv_h: 0.015
+hsv_s: 0.7
+hsv_v: 0.4
+degrees: 0.0
+translate: 0.1
+scale: 0.5
+shear: 0.0
+perspective: 0.0
+flipud: 0.0
+fliplr: 0.5
+mosaic: 1.0
+mixup: 0.0
+copy_paste: 0.0
+cfg: null
+tracker: botsort.yaml
+save_dir: small_grabage\CBAM-P2
diff --git a/small_grabage/CBAM-P2/events.out.tfevents.1700629108.Harry4090.77596.1 b/small_grabage/CBAM-P2/events.out.tfevents.1700629108.Harry4090.77596.1
diff --git a/small_grabage/CBAM-P2/labels.jpg b/small_grabage/CBAM-P2/labels.jpg
diff --git a/small_grabage/CBAM-P2/labels_correlogram.jpg b/small_grabage/CBAM-P2/labels_correlogram.jpg
diff --git a/small_grabage/CBAM-P2/results.csv b/small_grabage/CBAM-P2/results.csv
diff --git a/small_grabage/CBAM-P2/train_batch0.jpg b/small_grabage/CBAM-P2/train_batch0.jpg
diff --git a/small_grabage/CBAM-P2/train_batch1.jpg b/small_grabage/CBAM-P2/train_batch1.jpg
diff --git a/small_grabage/CBAM-P2/train_batch2.jpg b/small_grabage/CBAM-P2/train_batch2.jpg
diff --git a/small_grabage/CBAM-P2/weights/best.pt b/small_grabage/CBAM-P2/weights/best.pt
diff --git a/small_grabage/CBAM-P2/weights/last.pt b/small_grabage/CBAM-P2/weights/last.pt
diff --git a/small_grabage/CBAM/F1_curve.png b/small_grabage/CBAM/F1_curve.png
diff --git a/small_grabage/CBAM/PR_curve.png b/small_grabage/CBAM/PR_curve.png
diff --git a/small_grabage/CBAM/P_curve.png b/small_grabage/CBAM/P_curve.png
diff --git a/small_grabage/CBAM/R_curve.png b/small_grabage/CBAM/R_curve.png
diff --git a/small_grabage/CBAM/args.yaml b/small_grabage/CBAM/args.yaml
@@ -0,0 +1,98 @@
+task: detect
+mode: train
+model: yolov8n-CBAM.yaml
+data: data.yaml
+epochs: 1000
+patience: 50
+batch: 16
+imgsz: 256
+save: true
+save_period: -1
+cache: false
+device: null
+workers: 8
+project: small_grabage
+name: CBAM
+exist_ok: false
+pretrained: true
+optimizer: auto
+verbose: true
+seed: 0
+deterministic: true
+single_cls: false
+rect: false
+cos_lr: false
+close_mosaic: 10
+resume: null
+amp: true
+fraction: 1.0
+profile: false
+freeze: null
+overlap_mask: true
+mask_ratio: 4
+dropout: 0.0
+val: true
+split: val
+save_json: false
+save_hybrid: false
+conf: null
+iou: 0.7
+max_det: 300
+half: false
+dnn: false
+plots: true
+source: null
+show: false
+save_txt: false
+save_conf: false
+save_crop: false
+show_labels: true
+show_conf: true
+vid_stride: 1
+stream_buffer: false
+line_width: null
+visualize: false
+augment: false
+agnostic_nms: false
+classes: null
+retina_masks: false
+boxes: true
+format: torchscript
+keras: false
+optimize: false
+int8: false
+dynamic: false
+simplify: false
+opset: null
+workspace: 4
+nms: false
+lr0: 0.01
+lrf: 0.01
+momentum: 0.937
+weight_decay: 0.0005
+warmup_epochs: 3.0
+warmup_momentum: 0.8
+warmup_bias_lr: 0.1
+box: 7.5
+cls: 0.5
+dfl: 1.5
+pose: 12.0
+kobj: 1.0
+label_smoothing: 0.0
+nbs: 64
+hsv_h: 0.015
+hsv_s: 0.7
+hsv_v: 0.4
+degrees: 0.0
+translate: 0.1
+scale: 0.5
+shear: 0.0
+perspective: 0.0
+flipud: 0.0
+fliplr: 0.5
+mosaic: 1.0
+mixup: 0.0
+copy_paste: 0.0
+cfg: null
+tracker: botsort.yaml
+save_dir: small_grabage\CBAM
diff --git a/small_grabage/CBAM/confusion_matrix.png b/small_grabage/CBAM/confusion_matrix.png
diff --git a/small_grabage/CBAM/confusion_matrix_normalized.png b/small_grabage/CBAM/confusion_matrix_normalized.png
diff --git a/small_grabage/CBAM/events.out.tfevents.1700602530.Harry4090.77596.0 b/small_grabage/CBAM/events.out.tfevents.1700602530.Harry4090.77596.0
diff --git a/small_grabage/CBAM/labels.jpg b/small_grabage/CBAM/labels.jpg
diff --git a/small_grabage/CBAM/labels_correlogram.jpg b/small_grabage/CBAM/labels_correlogram.jpg
diff --git a/small_grabage/CBAM/results.csv b/small_grabage/CBAM/results.csv
diff --git a/small_grabage/CBAM/results.png b/small_grabage/CBAM/results.png
diff --git a/small_grabage/CBAM/train_batch0.jpg b/small_grabage/CBAM/train_batch0.jpg
diff --git a/small_grabage/CBAM/train_batch1.jpg b/small_grabage/CBAM/train_batch1.jpg
diff --git a/small_grabage/CBAM/train_batch2.jpg b/small_grabage/CBAM/train_batch2.jpg
diff --git a/small_grabage/CBAM/train_batch261360.jpg b/small_grabage/CBAM/train_batch261360.jpg
diff --git a/small_grabage/CBAM/train_batch261361.jpg b/small_grabage/CBAM/train_batch261361.jpg
diff --git a/small_grabage/CBAM/train_batch261362.jpg b/small_grabage/CBAM/train_batch261362.jpg
diff --git a/small_grabage/CBAM/val_batch0_labels.jpg b/small_grabage/CBAM/val_batch0_labels.jpg
diff --git a/small_grabage/CBAM/val_batch0_pred.jpg b/small_grabage/CBAM/val_batch0_pred.jpg
diff --git a/small_grabage/CBAM/val_batch1_labels.jpg b/small_grabage/CBAM/val_batch1_labels.jpg
diff --git a/small_grabage/CBAM/val_batch1_pred.jpg b/small_grabage/CBAM/val_batch1_pred.jpg
diff --git a/small_grabage/CBAM/val_batch2_labels.jpg b/small_grabage/CBAM/val_batch2_labels.jpg
diff --git a/small_grabage/CBAM/val_batch2_pred.jpg b/small_grabage/CBAM/val_batch2_pred.jpg
diff --git a/small_grabage/CBAM/weights/best.pt b/small_grabage/CBAM/weights/best.pt
diff --git a/small_grabage/CBAM/weights/best.torchscript b/small_grabage/CBAM/weights/best.torchscript
diff --git a/small_grabage/CBAM/weights/last.pt b/small_grabage/CBAM/weights/last.pt
diff --git a/small_grabage/CBAM2/F1_curve.png b/small_grabage/CBAM2/F1_curve.png
diff --git a/small_grabage/CBAM2/PR_curve.png b/small_grabage/CBAM2/PR_curve.png
diff --git a/small_grabage/CBAM2/P_curve.png b/small_grabage/CBAM2/P_curve.png
diff --git a/small_grabage/CBAM2/R_curve.png b/small_grabage/CBAM2/R_curve.png
diff --git a/small_grabage/CBAM2/confusion_matrix.png b/small_grabage/CBAM2/confusion_matrix.png
diff --git a/small_grabage/CBAM2/confusion_matrix_normalized.png b/small_grabage/CBAM2/confusion_matrix_normalized.png
diff --git a/small_grabage/CBAM2/val_batch0_labels.jpg b/small_grabage/CBAM2/val_batch0_labels.jpg
diff --git a/small_grabage/CBAM2/val_batch0_pred.jpg b/small_grabage/CBAM2/val_batch0_pred.jpg
diff --git a/small_grabage/CBAM2/val_batch1_labels.jpg b/small_grabage/CBAM2/val_batch1_labels.jpg
diff --git a/small_grabage/CBAM2/val_batch1_pred.jpg b/small_grabage/CBAM2/val_batch1_pred.jpg
diff --git a/small_grabage/CBAM2/val_batch2_labels.jpg b/small_grabage/CBAM2/val_batch2_labels.jpg
diff --git a/small_grabage/CBAM2/val_batch2_pred.jpg b/small_grabage/CBAM2/val_batch2_pred.jpg
diff --git a/ultralytics/ultralytics/cfg/models/v8/yolov8-CBAM-p2.yaml b/ultralytics/ultralytics/cfg/models/v8/yolov8-CBAM-p2.yaml
@@ -0,0 +1,56 @@
+# Ultralytics YOLO 🚀, AGPL-3.0 license
+# YOLOv8 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
+
+# Parameters
+nc: 80  # number of classes
+scales: # model compound scaling constants, i.e. 'model=yolov8n.yaml' will call yolov8.yaml with scale 'n'
+  # [depth, width, max_channels]
+  n: [0.33, 0.25, 1024]  # YOLOv8n summary: 225 layers,  3157200 parameters,  3157184 gradients,   8.9 GFLOPs
+  s: [0.33, 0.50, 1024]  # YOLOv8s summary: 225 layers, 11166560 parameters, 11166544 gradients,  28.8 GFLOPs
+  m: [0.67, 0.75, 768]   # YOLOv8m summary: 295 layers, 25902640 parameters, 25902624 gradients,  79.3 GFLOPs
+  l: [1.00, 1.00, 512]   # YOLOv8l summary: 365 layers, 43691520 parameters, 43691504 gradients, 165.7 GFLOPs
+  x: [1.00, 1.25, 512]   # YOLOv8x summary: 365 layers, 68229648 parameters, 68229632 gradients, 258.5 GFLOPs
+
+# YOLOv8.0n backbone
+backbone:
+  # [from, repeats, module, args]
+  - [-1, 1, Conv, [64, 3, 2]]  # 0-P1/2
+  - [-1, 1, Conv, [128, 3, 2]]  # 1-P2/4
+  - [-1, 3, C2f, [128, True]]
+  - [-1, 1, Conv, [256, 3, 2]]  # 3-P3/8
+  - [-1, 6, C2f, [256, True]]
+  - [-1, 1, Conv, [512, 3, 2]]  # 5-P4/16
+  - [-1, 6, C2f, [512, True]]
+  - [-1, 1, Conv, [1024, 3, 2]]  # 7-P5/32
+  - [-1, 3, C2f, [1024, True]]
+  - [-1, 3, CBAMBottleneck, [1024, 3]]
+  - [-1, 1, SPPF, [1024, 5]]  # 10
+
+# YOLOv8.0n head
+head:
+  - [-1, 1, inv, [1024, 1, 1]]
+  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
+  - [[-1, 6], 1, Concat, [1]]  # cat backbone P4
+  - [-1, 3, C2f, [512]]  # 14
+
+  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
+  - [[-1, 4], 1, Concat, [1]]  # cat backbone P3
+  - [-1, 3, C2f, [256]]  # 17 (P3/8-small)
+
+  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
+  - [[-1, 2], 1, Concat, [1]]  # cat backbone P2
+  - [-1, 3, C2f, [128]]  # 20 (P2/4-xsmall)
+
+  - [-1, 1, Conv, [128, 3, 2]]
+  - [[-1, 15], 1, Concat, [1]]  # cat head P3
+  - [-1, 3, C2f, [256]]  # 23 (P3/8-small)
+
+  - [-1, 1, Conv, [256, 3, 2]]
+  - [[-1, 12], 1, Concat, [1]]  # cat head P4
+  - [-1, 3, C2f, [512]]  # 26 (P4/16-medium)
+
+  - [-1, 1, Conv, [512, 3, 2]]
+  - [[-1, 9], 1, Concat, [1]]  # cat head P5
+  - [-1, 3, C2f, [1024]]  # 29 (P5/32-large)
+
+  - [[20, 23, 26, 29], 1, Detect, [nc]]  # Detect(P2, P3, P4, P5)
diff --git a/ultralytics/ultralytics/cfg/models/v8/yolov8-CBAM.yaml b/ultralytics/ultralytics/cfg/models/v8/yolov8-CBAM.yaml
@@ -0,0 +1,48 @@
+# Ultralytics YOLO 🚀, AGPL-3.0 license
+# YOLOv8 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
+
+# Parameters
+nc: 80  # number of classes
+scales: # model compound scaling constants, i.e. 'model=yolov8n.yaml' will call yolov8.yaml with scale 'n'
+  # [depth, width, max_channels]
+  n: [0.33, 0.25, 1024]  # YOLOv8n summary: 225 layers,  3157200 parameters,  3157184 gradients,   8.9 GFLOPs
+  s: [0.33, 0.50, 1024]  # YOLOv8s summary: 225 layers, 11166560 parameters, 11166544 gradients,  28.8 GFLOPs
+  m: [0.67, 0.75, 768]   # YOLOv8m summary: 295 layers, 25902640 parameters, 25902624 gradients,  79.3 GFLOPs
+  l: [1.00, 1.00, 512]   # YOLOv8l summary: 365 layers, 43691520 parameters, 43691504 gradients, 165.7 GFLOPs
+  x: [1.00, 1.25, 512]   # YOLOv8x summary: 365 layers, 68229648 parameters, 68229632 gradients, 258.5 GFLOPs
+
+# YOLOv8.0n backbone
+backbone:
+  # [from, repeats, module, args]
+  - [-1, 1, Conv, [64, 3, 2]]  # 0-P1/2
+  - [-1, 1, Conv, [128, 3, 2]]  # 1-P2/4
+  - [-1, 3, C2f, [128, True]]
+  - [-1, 1, Conv, [256, 3, 2]]  # 3-P3/8
+  - [-1, 6, C2f, [256, True]]
+  - [-1, 1, Conv, [512, 3, 2]]  # 5-P4/16
+  - [-1, 6, C2f, [512, True]]
+  - [-1, 1, Conv, [1024, 3, 2]]  # 7-P5/32
+  - [-1, 3, C2f, [1024, True]]
+  - [-1, 3, CBAMBottleneck, [1024, 3]]
+  - [-1, 1, SPPF, [1024, 5]]  # 10
+
+# YOLOv8.0n head
+head:
+  - [-1, 1, inv, [1024, 1, 1]]
+  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
+  - [[-1, 6], 1, Concat, [1]]  # cat backbone P4
+  - [-1, 3, C2f, [512]]  # 13
+
+  - [-1, 1, nn.Upsample, [None, 2, 'nearest']]
+  - [[-1, 4], 1, Concat, [1]]  # cat backbone P3
+  - [-1, 3, C2f, [256]]  # 16 (P3/8-small)
+
+  - [-1, 1, Conv, [256, 3, 2]]
+  - [[-1, 12], 1, Concat, [1]]  # cat head P4
+  - [-1, 3, C2f, [512]]  # 19 (P4/16-medium)
+
+  - [-1, 1, Conv, [512, 3, 2]]
+  - [[-1, 9], 1, Concat, [1]]  # cat head P5
+  - [-1, 3, C2f, [1024]]  # 22 (P5/32-large)
+
+  - [[16, 19, 22], 1, Detect, [nc]]  # Detect(P3, P4, P5)
diff --git a/ultralytics/ultralytics/cfg/models/v8/yolov8-std.yaml b/ultralytics/ultralytics/cfg/models/v8/yolov8-std.yaml
diff --git a/ultralytics/ultralytics/nn/modules/__init__.py b/ultralytics/ultralytics/nn/modules/__init__.py
@@ -20,7 +20,7 @@
 from .block import (C1, C2, C3, C3TR, DFL, SPP, SPPF, Bottleneck, BottleneckCSP, C2f, C3Ghost, C3x, GhostBottleneck,
                     HGBlock, HGStem, Proto, RepC3,space_to_depth)
 from .conv import (CBAM, ChannelAttention, Concat, Conv, Conv2, ConvTranspose, DWConv, DWConvTranspose2d, Focus,
-                   GhostConv, LightConv, RepConv, SpatialAttention)
+                   GhostConv, LightConv, RepConv, SpatialAttention,inv,CBAMBottleneck)
 from .head import Classify, Detect, Pose, RTDETRDecoder, Segment
 from .transformer import (AIFI, MLP, DeformableTransformerDecoder, DeformableTransformerDecoderLayer, LayerNorm2d,
                           MLPBlock, MSDeformAttn, TransformerBlock, TransformerEncoderLayer, TransformerLayer)
@@ -30,4 +30,4 @@
            'TransformerBlock', 'MLPBlock', 'LayerNorm2d', 'DFL', 'HGBlock', 'HGStem', 'SPP', 'SPPF', 'C1', 'C2', 'C3',
            'C2f', 'C3x', 'C3TR', 'C3Ghost', 'GhostBottleneck', 'Bottleneck', 'BottleneckCSP', 'Proto', 'Detect',
            'Segment', 'Pose', 'Classify', 'TransformerEncoderLayer', 'RepC3', 'RTDETRDecoder', 'AIFI',
-           'DeformableTransformerDecoder', 'DeformableTransformerDecoderLayer', 'MSDeformAttn', 'MLP','space_to_depth')
+           'DeformableTransformerDecoder', 'DeformableTransformerDecoderLayer', 'MSDeformAttn', 'MLP','space_to_depth','inv','CBAMBottleneck')