create: 注释

data/coco128.yaml

@@ -8,8 +8,11 @@
 
 
 # Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
+#数据集存放目录
 path: ../datasets/coco128  # dataset root dir
+#训练集存放目录
 train: images/train2017  # train images (relative to 'path') 128 images
+#验证集存放目录
 val: images/train2017  # val images (relative to 'path') 128 images
 test:  # test images (optional)
 

models/yolo.py

@@ -163,7 +163,7 @@ class BaseModel(nn.Module):
 
 
 class DetectionModel(BaseModel):
-    # YOLOv5 detection model
+    # YOLOv5 detection model 模型初始化
     def __init__(self, cfg='yolov5s.yaml', ch=3, nc=None, anchors=None):  # model, input channels, number of classes
         super().__init__()
         if isinstance(cfg, dict):
@@ -357,6 +357,7 @@ def parse_model(d, ch):  # model_dict, input_channels(3)
 
 
 if __name__ == '__main__':
+    # 定义参数信息
     parser = argparse.ArgumentParser()
     parser.add_argument('--cfg', type=str, default='yolov5s.yaml', help='model.yaml')
     parser.add_argument('--batch-size', type=int, default=1, help='total batch size for all GPUs')

utils/augmentations.py

@@ -110,21 +110,22 @@ def replicate(im, labels):
 
 def letterbox(im, new_shape=(640, 640), color=(114, 114, 114), auto=True, scaleFill=False, scaleup=True, stride=32):
     # Resize and pad image while meeting stride-multiple constraints
-    shape = im.shape[:2]  # current shape [height, width]
-    if isinstance(new_shape, int):
+    shape = im.shape[:2]  # current shape [height, width] 原图的高和宽[h,w] (1080,810)
+    if isinstance(new_shape, int): # 判断是不是int型的
         new_shape = (new_shape, new_shape)
 
     # Scale ratio (new / old)
+    # 按照新宽高/旧宽高 小的那一个，进行缩放
     r = min(new_shape[0] / shape[0], new_shape[1] / shape[1])
     if not scaleup:  # only scale down, do not scale up (for better val mAP)
         r = min(r, 1.0)
 
     # Compute padding
     ratio = r, r  # width, height ratios
-    new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r))
-    dw, dh = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]  # wh padding
-    if auto:  # minimum rectangle
-        dw, dh = np.mod(dw, stride), np.mod(dh, stride)  # wh padding
+    new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r)) # 求出缩放后的宽高 [width,height] 新图的尺寸 (480,640)
+    dw, dh = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]  # wh padding 160,0
+    if auto:  # minimum rectangle 最小矩形
+        dw, dh = np.mod(dw, stride), np.mod(dh, stride)  # wh padding 如果满足32的倍数，自动填充，例如480为32倍数，所以padding就为0，dw=0,dh=0
     elif scaleFill:  # stretch
         dw, dh = 0.0, 0.0
         new_unpad = (new_shape[1], new_shape[0])

utils/dataloaders.py

@@ -238,18 +238,18 @@ class LoadScreenshots:
 
 class LoadImages:
     # YOLOv5 image/video dataloader, i.e. `python detect.py --source image.jpg/vid.mp4`
-    def __init__(self, path, img_size=640, stride=32, auto=True, transforms=None, vid_stride=1):
+    def __init__(self, path, img_size=640, stride=32, auto=True, transforms=None, vid_stride=1): # path：“data\\images\\bus.jpg”,img_size:[640,640]
         if isinstance(path, str) and Path(path).suffix == '.txt':  # *.txt file with img/vid/dir on each line
             path = Path(path).read_text().rsplit()
         files = []
         for p in sorted(path) if isinstance(path, (list, tuple)) else [path]:
-            p = str(Path(p).resolve())
-            if '*' in p:
+            p = str(Path(p).resolve()) # 由相对路径得到绝对路径
+            if '*' in p: # 判断路径是否带"*"号
                 files.extend(sorted(glob.glob(p, recursive=True)))  # glob
-            elif os.path.isdir(p):
+            elif os.path.isdir(p): # 判断路径是否是文件夹
                 files.extend(sorted(glob.glob(os.path.join(p, '*.*'))))  # dir
-            elif os.path.isfile(p):
-                files.append(p)  # files
+            elif os.path.isfile(p): # 判断是不是文件
+                files.append(p)  # files, 变成列表形式
             else:
                 raise FileNotFoundError(f'{p} does not exist')
 
@@ -306,12 +306,13 @@ class LoadImages:
             self.count += 1
             im0 = cv2.imread(path)  # BGR
             assert im0 is not None, f'Image Not Found {path}'
-            s = f'image {self.count}/{self.nf} {path}: '
+            s = f'image {self.count}/{self.nf} {path}: ' # 打印图片输出到第几张了
 
         if self.transforms:
             im = self.transforms(im0)  # transforms
         else:
-            im = letterbox(im0, self.img_size, stride=self.stride, auto=self.auto)[0]  # padded resize
+            # 将原图变成特定大小的图片,resize
+            im = letterbox(im0, self.img_size, stride=self.stride, auto=self.auto)[0]  # padded resize (640,480,3)
             im = im.transpose((2, 0, 1))[::-1]  # HWC to CHW, BGR to RGB
             im = np.ascontiguousarray(im)  # contiguous
 

utils/general.py

@@ -130,12 +130,12 @@ def set_logging(name=LOGGING_NAME, verbose=True):
             name: {
                 'class': 'logging.StreamHandler',
                 'formatter': name,
-                'level': level,}},
+                'level': level, }},
         'loggers': {
             name: {
                 'level': level,
                 'handlers': [name],
-                'propagate': False,}}})
+                'propagate': False, }}})
 
 
 set_logging(LOGGING_NAME)  # run before defining LOGGER

utils/torch_utils.py

@@ -317,7 +317,9 @@ def copy_attr(a, b, include=(), exclude=()):
 
 def smart_optimizer(model, name='Adam', lr=0.001, momentum=0.9, decay=1e-5):
     # YOLOv5 3-param group optimizer: 0) weights with decay, 1) weights no decay, 2) biases no decay
+    # 把模型每一层的参数，划分到三个组中
     g = [], [], []  # optimizer parameter groups
+    # g[0]存所有卷积层的w参数，g[1] bn层的w，g[2]所有层的偏置项b
     bn = tuple(v for k, v in nn.__dict__.items() if 'Norm' in k)  # normalization layers, i.e. BatchNorm2d()
     for v in model.modules():
         for p_name, p in v.named_parameters(recurse=0):
@@ -327,7 +329,7 @@ def smart_optimizer(model, name='Adam', lr=0.001, momentum=0.9, decay=1e-5):
                 g[1].append(p)
             else:
                 g[0].append(p)  # weight (with decay)
-
+    # 判断训练过程中所使用的优化器的类型
     if name == 'Adam':
         optimizer = torch.optim.Adam(g[2], lr=lr, betas=(momentum, 0.999))  # adjust beta1 to momentum
     elif name == 'AdamW':
@@ -335,11 +337,12 @@ def smart_optimizer(model, name='Adam', lr=0.001, momentum=0.9, decay=1e-5):
     elif name == 'RMSProp':
         optimizer = torch.optim.RMSprop(g[2], lr=lr, momentum=momentum)
     elif name == 'SGD':
-        optimizer = torch.optim.SGD(g[2], lr=lr, momentum=momentum, nesterov=True)
+        # 默认随机梯度下降法
+        optimizer = torch.optim.SGD(g[2], lr=lr, momentum=momentum, nesterov=True) # lr 学习率， momentum 动量值
     else:
         raise NotImplementedError(f'Optimizer {name} not implemented.')
 
-    optimizer.add_param_group({'params': g[0], 'weight_decay': decay})  # add g0 with weight_decay
+    optimizer.add_param_group({'params': g[0], 'weight_decay': decay})  # add g0 with weight_decay  weight_decay标明要对卷积层的w进行权重衰减
     optimizer.add_param_group({'params': g[1], 'weight_decay': 0.0})  # add g1 (BatchNorm2d weights)
     LOGGER.info(f"{colorstr('optimizer:')} {type(optimizer).__name__}(lr={lr}) with parameter groups "
                 f'{len(g[1])} weight(decay=0.0), {len(g[0])} weight(decay={decay}), {len(g[2])} bias')
@@ -369,10 +372,12 @@ def smart_resume(ckpt, optimizer, ema=None, weights='yolov5s.pt', epochs=300, re
         ema.ema.load_state_dict(ckpt['ema'].float().state_dict())  # EMA
         ema.updates = ckpt['updates']
     if resume:
+        # 如果resume有值，且开始轮数大于0则接着训练，否则，抛出异常
         assert start_epoch > 0, f'{weights} training to {epochs} epochs is finished, nothing to resume.\n' \
                                 f"Start a new training without --resume, i.e. 'python train.py --weights {weights}'"
         LOGGER.info(f'Resuming training from {weights} from epoch {start_epoch} to {epochs} total epochs')
     if epochs < start_epoch:
+        # 如果开始轮数大于epochs，会提示已训练了多少轮，接下来会用使用的轮数进行微调
         LOGGER.info(f"{weights} has been trained for {ckpt['epoch']} epochs. Fine-tuning for {epochs} more epochs.")
         epochs += ckpt['epoch']  # finetune additional epochs
     return best_fitness, start_epoch, epochs