yolov26改进 | 添加注意力机制篇 | 2026医学最新注意力机制AFIA抑制图像噪声环境影响（全网独家首发）

​

import torch import torch.nn as nn import torch.nn.functional as F __all__ = ['AFIA', 'C2PSA_AFIA'] class ChannelShuffle(nn.Module): def __init__(self, groups): super().__init__() self.groups = groups def forward(self, x): b, c, h, w = x.size() channels_per_group = c // self.groups # reshape x = x.view(b, self.groups, channels_per_group, h, w) # transpose x = x.transpose(1, 2).contiguous() # flatten x = x.view(b, -1, h, w) return x class AFIA(nn.Module): def __init__(self, in_channels, groups=4): super().__init__() self.layer_norm = nn.LayerNorm(in_channels) # Assuming layer norm is applied over channels self.conv1x1_q = nn.Conv2d(in_channels, in_channels, kernel_size=1) self.deconv3x3_q = nn.ConvTranspose2d(in_channels, in_channels, kernel_size=3, stride=1, padding=1) self.conv1x1_k = nn.Conv2d(in_channels, in_channels, kernel_size=1) self.deconv3x3_k = nn.ConvTranspose2d(in_channels, in_channels, kernel_size=3, stride=1, padding=1) self.conv1x1_v = nn.Conv2d(in_channels, in_channels, kernel_size=1) self.deconv3x3_v = nn.ConvTranspose2d(in_channels, in_channels, kernel_size=3, stride=1, padding=1) self.channel_shuffle = ChannelShuffle(groups) self.stacked_conv = nn.Sequential( nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1), nn.ReLU(inplace=True), nn.Conv2d(in_channels, in_channels, kernel_size=3, padding=1), ) self.conv1x1_out = nn.Conv2d(in_channels, in_channels, kernel_size=1) self.layer_norm2 = nn.LayerNorm(in_channels) def depthwise_spatial_attention(self, Q, K): """Calculates the Depthwise Spatial Attention Map between Q and K. Args: Q: Tensor of shape (B, C, H, W) K: Tensor of shape (B, C, H, W) Returns: A spatial attention map with shape (B, H, W) """ B, C, H, W = Q.shape Q = Q.permute(0, 2, 3, 1).reshape(B, H * W, C) # B, HW, C K = K.permute(0, 2, 3, 1).reshape(B, H * W, C) # B, HW, C attention_map = torch.matmul(Q, K.transpose(1, 2)) # B, HW, HW attention_map = F.softmax(attention_map / (C ** 0.5), dim=-1) # B, HW, HW return attention_map def spatial_wise_self_attention(self, V, attention_map): """ Apply spatial-wise self-attention to value feature map Args: V: Tensor of shape (B, C, H, W) attention_map: Attention map with shape (B, H, W, HW) Return: weighted V """ B, C, H, W = V.shape V = V.permute(0, 2, 3, 1).reshape(B, H * W, C) # B, HW, C weighted_V = torch.matmul(attention_map, V) weighted_V = weighted_V.reshape(B, H, W, C).permute(0, 3, 1, 2) return weighted_V def forward(self, x): residual = x x = x.permute(0, 2, 3, 1) # B,H,W,C x = self.layer_norm(x).permute(0, 3, 1, 2) # B,C,H,W # Q branch Q = self.conv1x1_q(x) Q = self.deconv3x3_q(Q) # K branch K = self.conv1x1_k(x) K = self.deconv3x3_k(K) # V branch V = self.conv1x1_v(x) V = self.deconv3x3_v(V) # Channel Shuffle x = self.channel_shuffle(x) # DSA attention_map = self.depthwise_spatial_attention(Q, K) # SSA out = self.spatial_wise_self_attention(V, attention_map) # Apply weights out = F.relu(out) # 1x1 Conv out out = self.conv1x1_out(out) # Add and output out = out + self.stacked_conv(x) out = out + residual out = out.permute(0, 2, 3, 1) out = self.layer_norm2(out).permute(0, 3, 1, 2) return out def autopad(k, p=None, d=1): # kernel, padding, dilation """Pad to 'same' shape outputs.""" if d > 1: k = d * (k - 1) + 1 if isinstance(k, int) else [d * (x - 1) + 1 for x in k] # actual kernel-size if p is None: p = k // 2 if isinstance(k, int) else [x // 2 for x in k] # auto-pad return p class Conv(nn.Module): """Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation).""" default_act = nn.SiLU() # default activation def __init__(self, c1, c2, k=1, s=1, p=None, g=1, d=1, act=True): """Initialize Conv layer with given arguments including activation.""" super().__init__() self.conv = nn.Conv2d(c1, c2, k, s, autopad(k, p, d), groups=g, dilation=d, bias=False) self.bn = nn.BatchNorm2d(c2) self.act = self.default_act if act is True else act if isinstance(act, nn.Module) else nn.Identity() def forward(self, x): """Apply convolution, batch normalization and activation to input tensor.""" return self.act(self.bn(self.conv(x))) def forward_fuse(self, x): """Perform transposed convolution of 2D data.""" return self.act(self.conv(x)) class PSABlock(nn.Module): def __init__(self, c, attn_ratio=0.5, num_heads=4, shortcut=True) -> None: """Initializes the PSABlock with attention and feed-forward layers for enhanced feature extraction.""" super().__init__() self.attn = AFIA(c, 4) self.ffn = nn.Sequential(Conv(c, c * 2, 1), Conv(c * 2, c, 1, act=False)) self.add = shortcut def forward(self, x): """Executes a forward pass through PSABlock, applying attention and feed-forward layers to the input tensor.""" x = x + self.attn(x) if self.add else self.attn(x) x = x + self.ffn(x) if self.add else self.ffn(x) return x class C2PSA_AFIA(nn.Module): """ C2PSA module with attention mechanism for enhanced feature extraction and processing. This module implements a convolutional block with attention mechanisms to enhance feature extraction and processing capabilities. It includes a series of PSABlock modules for self-attention and feed-forward operations. Attributes: c (int): Number of hidden channels. cv1 (Conv): 1x1 convolution layer to reduce the number of input channels to 2*c. cv2 (Conv): 1x1 convolution layer to reduce the number of output channels to c. m (nn.Sequential): Sequential container of PSABlock modules for attention and feed-forward operations. Methods: forward: Performs a forward pass through the C2PSA module, applying attention and feed-forward operations. Notes: This module essentially is the same as PSA module, but refactored to allow stacking more PSABlock modules. """ def __init__(self, c1, c2, n=1, e=0.5): """Initializes the C2PSA module with specified input/output channels, number of layers, and expansion ratio.""" super().__init__() assert c1 == c2 self.c = int(c1 * e) self.cv1 = Conv(c1, 2 * self.c, 1, 1) self.cv2 = Conv(2 * self.c, c1, 1) self.m = nn.Sequential(*(PSABlock(self.c, attn_ratio=0.5, num_heads=self.c // 64) for _ in range(n))) def forward(self, x): """Processes the input tensor 'x' through a series of PSA blocks and returns the transformed tensor.""" a, b = self.cv1(x).split((self.c, self.c), dim=1) b = self.m(b) return self.cv2(torch.cat((a, b), 1)) if __name__ == "__main__": # Generating Sample image image_size = (1, 64, 240, 240) image = torch.rand(*image_size) # Model mobilenet_v1 = C2PSA_AFIA(64, 64) out = mobilenet_v1(image) print(out.size())

elif m in {此处填写本章代码的名字.}: c2 = ch[f] args = [c2, *args]

# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license # Ultralytics YOLO26 object detection model with P3/8 - P5/32 outputs # Model docs: https://docs.ultralytics.com/models/yolo26 # Task docs: https://docs.ultralytics.com/tasks/detect # Parameters nc: 80 # number of classes end2end: True # whether to use end-to-end mode reg_max: 1 # DFL bins scales: # model compound scaling constants, i.e. 'model=yolo26n.yaml' will call yolo26.yaml with scale 'n' # [depth, width, max_channels] n: [0.50, 0.25, 1024] # summary: 260 layers, 2,572,280 parameters, 2,572,280 gradients, 6.1 GFLOPs s: [0.50, 0.50, 1024] # summary: 260 layers, 10,009,784 parameters, 10,009,784 gradients, 22.8 GFLOPs m: [0.50, 1.00, 512] # summary: 280 layers, 21,896,248 parameters, 21,896,248 gradients, 75.4 GFLOPs l: [1.00, 1.00, 512] # summary: 392 layers, 26,299,704 parameters, 26,299,704 gradients, 93.8 GFLOPs x: [1.00, 1.50, 512] # summary: 392 layers, 58,993,368 parameters, 58,993,368 gradients, 209.5 GFLOPs # YOLO26n 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, 2, C3k2, [256, False, 0.25]] - [-1, 1, Conv, [256, 3, 2]] # 3-P3/8 - [-1, 2, C3k2, [512, False, 0.25]] - [-1, 1, Conv, [512, 3, 2]] # 5-P4/16 - [-1, 2, C3k2, [512, True]] - [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32 - [-1, 2, C3k2, [1024, True]] - [-1, 1, SPPF, [1024, 5, 3, True]] # 9 - [-1, 2, C2PSA_AFIA, [1024]] # 10 # YOLO26n head head: - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 6], 1, Concat, [1]] # cat backbone P4 - [-1, 2, C3k2, [512, True]] # 13 - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 4], 1, Concat, [1]] # cat backbone P3 - [-1, 2, C3k2, [256, True]] # 16 (P3/8-small) - [-1, 1, Conv, [256, 3, 2]] - [[-1, 13], 1, Concat, [1]] # cat head P4 - [-1, 2, C3k2, [512, True]] # 19 (P4/16-medium) - [-1, 1, Conv, [512, 3, 2]] - [[-1, 10], 1, Concat, [1]] # cat head P5 - [-1, 1, C3k2, [1024, True, 0.5, True]] # 22 (P5/32-large) - [[16, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5)

# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license # Ultralytics YOLO26 object detection model with P3/8 - P5/32 outputs # Model docs: https://docs.ultralytics.com/models/yolo26 # Task docs: https://docs.ultralytics.com/tasks/detect # Parameters nc: 80 # number of classes end2end: True # whether to use end-to-end mode reg_max: 1 # DFL bins scales: # model compound scaling constants, i.e. 'model=yolo26n.yaml' will call yolo26.yaml with scale 'n' # [depth, width, max_channels] n: [0.50, 0.25, 1024] # summary: 260 layers, 2,572,280 parameters, 2,572,280 gradients, 6.1 GFLOPs s: [0.50, 0.50, 1024] # summary: 260 layers, 10,009,784 parameters, 10,009,784 gradients, 22.8 GFLOPs m: [0.50, 1.00, 512] # summary: 280 layers, 21,896,248 parameters, 21,896,248 gradients, 75.4 GFLOPs l: [1.00, 1.00, 512] # summary: 392 layers, 26,299,704 parameters, 26,299,704 gradients, 93.8 GFLOPs x: [1.00, 1.50, 512] # summary: 392 layers, 58,993,368 parameters, 58,993,368 gradients, 209.5 GFLOPs # YOLO26n 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, 2, C3k2, [256, False, 0.25]] - [-1, 1, Conv, [256, 3, 2]] # 3-P3/8 - [-1, 2, C3k2, [512, False, 0.25]] - [-1, 1, Conv, [512, 3, 2]] # 5-P4/16 - [-1, 2, C3k2, [512, True]] - [-1, 1, Conv, [1024, 3, 2]] # 7-P5/32 - [-1, 2, C3k2, [1024, True]] - [-1, 1, SPPF, [1024, 5, 3, True]] # 9 - [-1, 2, C2PSA, [1024]] # 10 # YOLO26n head head: - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 6], 1, Concat, [1]] # cat backbone P4 - [-1, 2, C3k2, [512, True]] # 13 - [-1, 1, nn.Upsample, [None, 2, "nearest"]] - [[-1, 4], 1, Concat, [1]] # cat backbone P3 - [-1, 2, C3k2, [256, True]] # 16 (P3/8-small) - [-1, 1, Conv, [256, 3, 2]] - [[-1, 13], 1, Concat, [1]] # cat head P4 - [-1, 2, C3k2, [512, True]] # 19 (P4/16-medium) - [-1, 1, Conv, [512, 3, 2]] - [[-1, 10], 1, Concat, [1]] # cat head P5 - [-1, 1, C3k2, [1024, True, 0.5, True]] # 22 (P5/32-large) - [16, 1, AFIA, []] # 23 # - [19, 1, AFIA, []] # 24 # - [22, 1, AFIA, []] # 25 # 此处的使用说法注释: 其中上面的三个注意力机制目前仅使用了23层，如果你想使用24层那么就取消掉代码注释， # 并将下面检测头中的19改为24,如果想使用第25层注意力机制同理，将下面检测头中的22改为25即可。 # 此处用法比较复杂如过不会联系Snu77博主获取视频教程 - [[23, 19, 22], 1, Detect, [nc]] # Detect(P3, P4, P5)

import warnings warnings.filterwarnings('ignore') from ultralytics import YOLO if __name__ == '__main__': model = YOLO('模型配置文件地址,也就是5.1你保存到本地文件的地址') # 如何切换模型版本, 上面的ymal文件可以改为 yolo26s.yaml就是使用的26s, # 类似某个改进的yaml文件名称为yolo26-XXX.yaml那么如果想使用其它版本就把上面的名称改为yolo26l-XXX.yaml即可（改的是上面YOLO中间的名字不是配置文件的）！ # model.load('yolo26n.pt') # 是否加载预训练权重,科研不建议大家加载否则很难提升精度 model.train( data=r"数据集文件地址", # 如果大家任务是其它的'ultralytics/cfg/default.yaml'找到这里修改task可以改成detect, segment, classify, pose cache=False, imgsz=640, epochs=20, single_cls=False, # 是否是单类别检测 batch=16, close_mosaic=0, workers=0, device='0', optimizer='MuSGD', # using SGD/MuSGD # resume=, # 这里是填写last.pt地址 amp=True, # 如果出现训练损失为Nan可以关闭amp project='runs/train', name='exp', )