【YOLOv12多模态融合改进】在前期、中期、中后期、后期多模态融合中添加P2小目标检测层，完整步骤及代码
前言
主题： YOLOv12 的多模态融合改进中增加P2小目标检测层
方式： 分别在前期融合、中期融合、中-后期融合、后期融合中增加P2多模态融合检测层。
内容： 包含融合方式详解以及完整配置步骤，开箱即用，一键运行。
一、YOLOv12原始模型结构介绍
YOLOv12
原始模型结构如下：
# YOLOv12 🚀, AGPL-3.0 license
# YOLOv12 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
# CFG file for YOLOv12-turbo
# Parameters
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov12n.yaml' will call yolov12.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 497 layers, 2,553,904 parameters, 2,553,888 gradients, 6.2 GFLOPs
s: [0.50, 0.50, 1024] # summary: 497 layers, 9,127,424 parameters, 9,127,408 gradients, 19.7 GFLOPs
m: [0.50, 1.00, 512] # summary: 533 layers, 19,670,784 parameters, 19,670,768 gradients, 60.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 895 layers, 26,506,496 parameters, 26,506,480 gradients, 83.3 GFLOPs
x: [1.00, 1.50, 512] # summary: 895 layers, 59,414,176 parameters, 59,414,160 gradients, 185.9 GFLOPs
# YOLO12 backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, Conv, [64, 3, 2]] # 0-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 1-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 3-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 5-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 8
# YOLO12 head
head:
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 6], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 11
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 4], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 14
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 11], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 17
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 8], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 20 (P5/32-large)
- [[14, 17, 20], 1, Detect, [nc]] # Detect(P3, P4, P5)
二、有效特征层对应的检测头类别
2.1 P3/8 - small检测头
-
原始模型中的
P3/8特征层对应的检测头主要用于检测相对较小的目标。其特征图大小相对较大，空间分辨率较高。 -
适合检测尺寸大概在
8x8到32x32像素左右的目标。
2.2 P4/16 - medium检测头
-
这个检测头对应的
P4/16特征层经过了更多的下采样操作，相比P3/8特征图空间分辨率降低，但通道数增加，特征更抽象且有语义信息。 -
它主要用于检测中等大小的目标，尺寸范围大概在
32x32到64x64像素左右。
2.3 P5/32 - large检测头
-
P5/32是经过最多下采样操作得到的特征层，其空间分辨率最低，但语义信息最强、全局感受野最大。 -
该检测头适合检测较大尺寸的目标，一般是尺寸在
64x64像素以上的目标。
2.4 新添加针对小目标的检测头
-
新添加的检测头主要用于检测更小尺寸的目标。尺寸在
4x4到8x8像素左右的微小目标。
💡这是因为在目标检测任务中，随着目标尺寸的减小，需要更高分辨率的特征图来有效捕捉目标特征。新添加的检测头很可能是基于这样的考虑，通过一系列的卷积、上采样和拼接等操作生成适合微小目标检测的特征图，从而提高模型对微小目标的检测能力。
三、小目标检测头多模态融合方式
-
前期融合中，在网络输入阶段将多模态数据合并后，增加针对小目标的检测层。
-
中期融合中，在骨干网络中增加针对P2的多模态特征进行融合，以此引出小目标的检测层。
-
中-后期融合中，在颈部的FPN结构中，增加针对P2的多模态特征进行融合，以此引出小目标的检测层。
-
后期融合中，在检测头前增加P2多模态特征进行融合。
四、完整配置步骤
！！！ 私信获取的项目包就已经把相关的多模态输入、训练等改动都已经配好了，只需要新建模型yaml文件，粘贴对应的模型，进行训练即可。 项目包获取及使用教程可参考链接： 《YOLO系列模型的多模态项目》配置使用教程
在什么地方新建，n,s,m,l,x,用哪个版本按自己的需求来即可，和普通的训练步骤一致。
除了模型结构方面的改动，在yaml文件中还传入了一个通道数
ch: 6
表示传入的是双模态，6通道 ，前三个是可见光，后三个是红外。
在default.yaml中也配置了这个参数。
4.1 P2前期融合
# YOLOv12 🚀, AGPL-3.0 license
# YOLOv12 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
# CFG file for YOLOv12-turbo
# Parameters
ch: 6
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov12n.yaml' will call yolov12.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 497 layers, 2,553,904 parameters, 2,553,888 gradients, 6.2 GFLOPs
s: [0.50, 0.50, 1024] # summary: 497 layers, 9,127,424 parameters, 9,127,408 gradients, 19.7 GFLOPs
m: [0.50, 1.00, 512] # summary: 533 layers, 19,670,784 parameters, 19,670,768 gradients, 60.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 895 layers, 26,506,496 parameters, 26,506,480 gradients, 83.3 GFLOPs
x: [1.00, 1.50, 512] # summary: 895 layers, 59,414,176 parameters, 59,414,160 gradients, 185.9 GFLOPs
# YOLO12 backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, MF, [64]] # 0
- [-1, 1, Conv, [64, 3, 2]] # 1-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 2-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 4-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 6-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 8-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 9
# YOLO12 head
head:
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 7], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 12
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 5], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 15
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 3], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 18
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 15], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [256, False, -1]] # 21
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 12], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 24
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 9], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 27 (P5/32-large)
- [[18, 21, 24, 27], 1, Detect, [nc]] # Detect(P3, P4, P5)
4.2 P2中期融合
# YOLOv12 🚀, AGPL-3.0 license
# YOLOv12 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
# CFG file for YOLOv12-turbo
# Parameters
ch: 6
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov12n.yaml' will call yolov12.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 497 layers, 2,553,904 parameters, 2,553,888 gradients, 6.2 GFLOPs
s: [0.50, 0.50, 1024] # summary: 497 layers, 9,127,424 parameters, 9,127,408 gradients, 19.7 GFLOPs
m: [0.50, 1.00, 512] # summary: 533 layers, 19,670,784 parameters, 19,670,768 gradients, 60.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 895 layers, 26,506,496 parameters, 26,506,480 gradients, 83.3 GFLOPs
x: [1.00, 1.50, 512] # summary: 895 layers, 59,414,176 parameters, 59,414,160 gradients, 185.9 GFLOPs
# YOLO12 backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, IN, []] # 0
- [-1, 1, Multiin, [1]] # 1
- [-2, 1, Multiin, [2]] # 2
- [1, 1, Conv, [64, 3, 2]] # 3-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 4-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 6-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 8-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 10-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 11
- [2, 1, Conv, [64, 3, 2]] # 12-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 13-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 15-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 17-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 19-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 20
- [[5, 14], 1, Concat, [1]] # 21 cat backbone P3
- [[7, 16], 1, Concat, [1]] # 22 cat backbone P3
- [[9, 18], 1, Concat, [1]] # 23 cat backbone P4
- [[11, 20], 1, Concat, [1]] # 24 cat backbone P5
# YOLO12 head
head:
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 23], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 27
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 22], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 30
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 21], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 33
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 30], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [256, False, -1]] # 36
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 27], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 39
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 24], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 42 (P5/32-large)
- [[33, 36, 39, 42], 1, Detect, [nc]] # Detect(P3, P4, P5)
4.3 P2中-后期融合
# YOLOv12 🚀, AGPL-3.0 license
# YOLOv12 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
# CFG file for YOLOv12-turbo
# Parameters
ch: 6
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov12n.yaml' will call yolov12.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 497 layers, 2,553,904 parameters, 2,553,888 gradients, 6.2 GFLOPs
s: [0.50, 0.50, 1024] # summary: 497 layers, 9,127,424 parameters, 9,127,408 gradients, 19.7 GFLOPs
m: [0.50, 1.00, 512] # summary: 533 layers, 19,670,784 parameters, 19,670,768 gradients, 60.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 895 layers, 26,506,496 parameters, 26,506,480 gradients, 83.3 GFLOPs
x: [1.00, 1.50, 512] # summary: 895 layers, 59,414,176 parameters, 59,414,160 gradients, 185.9 GFLOPs
# YOLO12 backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, IN, []] # 0
- [-1, 1, Multiin, [1]] # 1
- [-2, 1, Multiin, [2]] # 2
- [1, 1, Conv, [64, 3, 2]] # 3-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 4-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 6-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 8-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 10-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 11
- [2, 1, Conv, [64, 3, 2]] # 12-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 13-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 15-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 17-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 19-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 20
# YOLO12 head
head:
- [11, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 9], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 23
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 7], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 26
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 5], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 29
- [20, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 18], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 32
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 16], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 35
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 14], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 38
- [[11, 20], 1, Concat, [1]] # cat head P5 39
- [[23, 32], 1, Concat, [1]] # cat head P5 40
- [[26, 35], 1, Concat, [1]] # cat head P5 41
- [[29, 38], 1, Concat, [1]] # cat head P5 42
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 41], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [256, False, -1]] # 45
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 40], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 48
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 39], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 51 (P5/32-large)
- [[42, 45, 48, 51], 1, Detect, [nc]] # Detect(P3, P4, P5)
4.4 P2后期融合
# YOLOv12 🚀, AGPL-3.0 license
# YOLOv12 object detection model with P3-P5 outputs. For Usage examples see https://docs.ultralytics.com/tasks/detect
# CFG file for YOLOv12-turbo
# Parameters
ch: 6
nc: 80 # number of classes
scales: # model compound scaling constants, i.e. 'model=yolov12n.yaml' will call yolov12.yaml with scale 'n'
# [depth, width, max_channels]
n: [0.50, 0.25, 1024] # summary: 497 layers, 2,553,904 parameters, 2,553,888 gradients, 6.2 GFLOPs
s: [0.50, 0.50, 1024] # summary: 497 layers, 9,127,424 parameters, 9,127,408 gradients, 19.7 GFLOPs
m: [0.50, 1.00, 512] # summary: 533 layers, 19,670,784 parameters, 19,670,768 gradients, 60.4 GFLOPs
l: [1.00, 1.00, 512] # summary: 895 layers, 26,506,496 parameters, 26,506,480 gradients, 83.3 GFLOPs
x: [1.00, 1.50, 512] # summary: 895 layers, 59,414,176 parameters, 59,414,160 gradients, 185.9 GFLOPs
# YOLO12 backbone
backbone:
# [from, repeats, module, args]
- [-1, 1, IN, []] # 0
- [-1, 1, Multiin, [1]] # 1
- [-2, 1, Multiin, [2]] # 2
- [1, 1, Conv, [64, 3, 2]] # 3-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 4-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 6-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 8-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 10-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 11
- [2, 1, Conv, [64, 3, 2]] # 12-P1/2
- [-1, 1, Conv, [128, 3, 2, 1, 2]] # 13-P2/4
- [-1, 2, C3k2, [256, False, 0.25]]
- [-1, 1, Conv, [256, 3, 2, 1, 4]] # 15-P3/8
- [-1, 2, C3k2, [512, False, 0.25]]
- [-1, 1, Conv, [512, 3, 2]] # 17-P4/16
- [-1, 4, A2C2f, [512, True, 4]]
- [-1, 1, Conv, [1024, 3, 2]] # 19-P5/32
- [-1, 4, A2C2f, [1024, True, 1]] # 20
# YOLO12 head
head:
- [11, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 9], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 23
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 7], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 26
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 5], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 29
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 26], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [256, False, -1]] # 32
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 23], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 35
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 11], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 38 (P5/32-large)
- [20, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 18], 1, Concat, [1]] # cat backbone P4
- [-1, 2, A2C2f, [512, False, -1]] # 41
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 16], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 44
- [-1, 1, nn.Upsample, [None, 2, "nearest"]]
- [[-1, 14], 1, Concat, [1]] # cat backbone P3
- [-1, 2, A2C2f, [256, False, -1]] # 47
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 44], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [256, False, -1]] # 50
- [-1, 1, Conv, [256, 3, 2]]
- [[-1, 41], 1, Concat, [1]] # cat head P4
- [-1, 2, A2C2f, [512, False, -1]] # 53
- [-1, 1, Conv, [512, 3, 2]]
- [[-1, 20], 1, Concat, [1]] # cat head P5
- [-1, 2, C3k2, [1024, True]] # 56 (P5/32-large)
- [[29, 47], 1, Concat, [1]] # cat head P5 57
- [[32, 50], 1, Concat, [1]] # cat head P5 58
- [[35, 53], 1, Concat, [1]] # cat head P5 59
- [[38, 56], 1, Concat, [1]] # cat head P5 60
- [[57, 58, 59, 60], 1, Detect, [nc]] # Detect(P3, P4, P5)
五、成功运行结果
前期融合结果： 可以看到输入的通道数为6，表明可见光图像和红外图像均输入到了模型中进行融合训练。
YOLOv12-early-p2 summary: 575 layers, 2,705,500 parameters, 2,705,484 gradients, 13.1 GFLOPs
from n params module arguments
0 -1 1 472 ultralytics.nn.AddModules.multimodal.MF [6, 16]
1 -1 1 2336 ultralytics.nn.modules.conv.Conv [16, 16, 3, 2]
2 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
3 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
4 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
5 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
6 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
7 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
8 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
9 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
10 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
11 [-1, 7] 1 0 ultralytics.nn.modules.conv.Concat [1]
12 -1 1 86912 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 128, 1, False, -1]
13 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
14 [-1, 5] 1 0 ultralytics.nn.modules.conv.Concat [1]
15 -1 1 24000 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 64, 1, False, -1]
16 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
17 [-1, 3] 1 0 ultralytics.nn.modules.conv.Concat [1]
18 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
19 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
20 [-1, 15] 1 0 ultralytics.nn.modules.conv.Concat [1]
21 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
22 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
23 [-1, 12] 1 0 ultralytics.nn.modules.conv.Concat [1]
24 -1 1 74624 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 128, 1, False, -1]
25 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
26 [-1, 9] 1 0 ultralytics.nn.modules.conv.Concat [1]
27 -1 1 378880 ultralytics.nn.modules.block.C3k2 [384, 256, 1, True]
28 [18, 21, 24, 27] 1 518932 ultralytics.nn.modules.head.Detect [1, [64, 64, 128, 256]]
YOLOv12-early-p2 summary: 575 layers, 2,705,500 parameters, 2,705,484 gradients, 13.1 GFLOPs
中期融合结果：
YOLOv12-mid-p2 summary: 810 layers, 4,157,716 parameters, 4,157,700 gradients, 15.5 GFLOPs
from n params module arguments
0 -1 1 0 ultralytics.nn.AddModules.multimodal.IN []
1 -1 1 0 ultralytics.nn.AddModules.multimodal.Multiin [1]
2 -2 1 0 ultralytics.nn.AddModules.multimodal.Multiin [2]
3 1 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
4 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
5 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
6 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
7 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
8 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
9 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
10 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
11 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
12 2 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
13 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
14 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
15 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
16 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
17 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
18 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
19 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
20 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
21 [5, 14] 1 0 ultralytics.nn.modules.conv.Concat [1]
22 [7, 16] 1 0 ultralytics.nn.modules.conv.Concat [1]
23 [9, 18] 1 0 ultralytics.nn.modules.conv.Concat [1]
24 [11, 20] 1 0 ultralytics.nn.modules.conv.Concat [1]
25 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
26 [-1, 23] 1 0 ultralytics.nn.modules.conv.Concat [1]
27 -1 1 111488 ultralytics.nn.AddModules.A2C2f.A2C2f [768, 128, 1, False, -1]
28 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
29 [-1, 22] 1 0 ultralytics.nn.modules.conv.Concat [1]
30 -1 1 28096 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 64, 1, False, -1]
31 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
32 [-1, 21] 1 0 ultralytics.nn.modules.conv.Concat [1]
33 -1 1 21952 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 64, 1, False, -1]
34 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
35 [-1, 30] 1 0 ultralytics.nn.modules.conv.Concat [1]
36 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
37 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
38 [-1, 27] 1 0 ultralytics.nn.modules.conv.Concat [1]
39 -1 1 74624 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 128, 1, False, -1]
40 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
41 [-1, 24] 1 0 ultralytics.nn.modules.conv.Concat [1]
42 -1 1 444416 ultralytics.nn.modules.block.C3k2 [640, 256, 1, True]
43 [33, 36, 39, 42] 1 518932 ultralytics.nn.modules.head.Detect [1, [64, 64, 128, 256]]
YOLOv12-mid-p2 summary: 810 layers, 4,157,716 parameters, 4,157,700 gradients, 15.5 GFLOPs
中-后期融合结果：
YOLOv12-mid-to-late-p2 summary: 915 layers, 4,436,692 parameters, 4,436,676 gradients, 21.0 GFLOPs
from n params module arguments
0 -1 1 0 ultralytics.nn.AddModules.multimodal.IN []
1 -1 1 0 ultralytics.nn.AddModules.multimodal.Multiin [1]
2 -2 1 0 ultralytics.nn.AddModules.multimodal.Multiin [2]
3 1 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
4 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
5 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
6 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
7 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
8 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
9 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
10 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
11 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
12 2 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
13 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
14 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
15 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
16 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
17 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
18 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
19 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
20 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
21 11 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
22 [-1, 9] 1 0 ultralytics.nn.modules.conv.Concat [1]
23 -1 1 86912 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 128, 1, False, -1]
24 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
25 [-1, 7] 1 0 ultralytics.nn.modules.conv.Concat [1]
26 -1 1 24000 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 64, 1, False, -1]
27 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
28 [-1, 5] 1 0 ultralytics.nn.modules.conv.Concat [1]
29 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
30 20 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
31 [-1, 18] 1 0 ultralytics.nn.modules.conv.Concat [1]
32 -1 1 86912 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 128, 1, False, -1]
33 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
34 [-1, 16] 1 0 ultralytics.nn.modules.conv.Concat [1]
35 -1 1 24000 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 64, 1, False, -1]
36 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
37 [-1, 14] 1 0 ultralytics.nn.modules.conv.Concat [1]
38 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
39 [11, 20] 1 0 ultralytics.nn.modules.conv.Concat [1]
40 [23, 32] 1 0 ultralytics.nn.modules.conv.Concat [1]
41 [26, 35] 1 0 ultralytics.nn.modules.conv.Concat [1]
42 [29, 38] 1 0 ultralytics.nn.modules.conv.Concat [1]
43 -1 1 73856 ultralytics.nn.modules.conv.Conv [128, 64, 3, 2]
44 [-1, 41] 1 0 ultralytics.nn.modules.conv.Concat [1]
45 -1 1 21952 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 64, 1, False, -1]
46 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
47 [-1, 40] 1 0 ultralytics.nn.modules.conv.Concat [1]
48 -1 1 82816 ultralytics.nn.AddModules.A2C2f.A2C2f [320, 128, 1, False, -1]
49 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
50 [-1, 39] 1 0 ultralytics.nn.modules.conv.Concat [1]
51 -1 1 444416 ultralytics.nn.modules.block.C3k2 [640, 256, 1, True]
52 [42, 45, 48, 51] 1 650708 ultralytics.nn.modules.head.Detect [1, [128, 64, 128, 256]]
YOLOv12-mid-to-late-p2 summary: 915 layers, 4,436,692 parameters, 4,436,676 gradients, 21.0 GFLOPs
后期融合结果：
YOLOv12-late-p2 summary: 1,026 layers, 5,339,476 parameters, 5,339,460 gradients, 23.0 GFLOPs
from n params module arguments
0 -1 1 0 ultralytics.nn.AddModules.multimodal.IN []
1 -1 1 0 ultralytics.nn.AddModules.multimodal.Multiin [1]
2 -2 1 0 ultralytics.nn.AddModules.multimodal.Multiin [2]
3 1 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
4 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
5 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
6 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
7 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
8 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
9 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
10 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
11 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
12 2 1 464 ultralytics.nn.modules.conv.Conv [3, 16, 3, 2]
13 -1 1 2368 ultralytics.nn.modules.conv.Conv [16, 32, 3, 2, 1, 2]
14 -1 1 6640 ultralytics.nn.modules.block.C3k2 [32, 64, 1, False, 0.25]
15 -1 1 9344 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2, 1, 4]
16 -1 1 26080 ultralytics.nn.modules.block.C3k2 [64, 128, 1, False, 0.25]
17 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
18 -1 2 180864 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 128, 2, True, 4]
19 -1 1 295424 ultralytics.nn.modules.conv.Conv [128, 256, 3, 2]
20 -1 2 689408 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 256, 2, True, 1]
21 11 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
22 [-1, 9] 1 0 ultralytics.nn.modules.conv.Concat [1]
23 -1 1 86912 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 128, 1, False, -1]
24 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
25 [-1, 7] 1 0 ultralytics.nn.modules.conv.Concat [1]
26 -1 1 24000 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 64, 1, False, -1]
27 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
28 [-1, 5] 1 0 ultralytics.nn.modules.conv.Concat [1]
29 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
30 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
31 [-1, 26] 1 0 ultralytics.nn.modules.conv.Concat [1]
32 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
33 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
34 [-1, 23] 1 0 ultralytics.nn.modules.conv.Concat [1]
35 -1 1 74624 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 128, 1, False, -1]
36 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
37 [-1, 11] 1 0 ultralytics.nn.modules.conv.Concat [1]
38 -1 1 378880 ultralytics.nn.modules.block.C3k2 [384, 256, 1, True]
39 20 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
40 [-1, 18] 1 0 ultralytics.nn.modules.conv.Concat [1]
41 -1 1 86912 ultralytics.nn.AddModules.A2C2f.A2C2f [384, 128, 1, False, -1]
42 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
43 [-1, 16] 1 0 ultralytics.nn.modules.conv.Concat [1]
44 -1 1 24000 ultralytics.nn.AddModules.A2C2f.A2C2f [256, 64, 1, False, -1]
45 -1 1 0 torch.nn.modules.upsampling.Upsample [None, 2, 'nearest']
46 [-1, 14] 1 0 ultralytics.nn.modules.conv.Concat [1]
47 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
48 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
49 [-1, 44] 1 0 ultralytics.nn.modules.conv.Concat [1]
50 -1 1 19904 ultralytics.nn.AddModules.A2C2f.A2C2f [128, 64, 1, False, -1]
51 -1 1 36992 ultralytics.nn.modules.conv.Conv [64, 64, 3, 2]
52 [-1, 41] 1 0 ultralytics.nn.modules.conv.Concat [1]
53 -1 1 74624 ultralytics.nn.AddModules.A2C2f.A2C2f [192, 128, 1, False, -1]
54 -1 1 147712 ultralytics.nn.modules.conv.Conv [128, 128, 3, 2]
55 [-1, 20] 1 0 ultralytics.nn.modules.conv.Concat [1]
56 -1 1 378880 ultralytics.nn.modules.block.C3k2 [384, 256, 1, True]
57 [29, 47] 1 0 ultralytics.nn.modules.conv.Concat [1]
58 [32, 50] 1 0 ultralytics.nn.modules.conv.Concat [1]
59 [35, 53] 1 0 ultralytics.nn.modules.conv.Concat [1]
60 [38, 56] 1 0 ultralytics.nn.modules.conv.Concat [1]
61 [57, 58, 59, 60] 1 971028 ultralytics.nn.modules.head.Detect [1, [128, 128, 256, 512]]
YOLOv12-late-p2 summary: 1,026 layers, 5,339,476 parameters, 5,339,460 gradients, 23.0 GFLOPs