diff --git a/.github/workflows/merge-main-into-prs.yml b/.github/workflows/merge-main-into-prs.yml
index eae8e00..05a0363 100644
--- a/.github/workflows/merge-main-into-prs.yml
+++ b/.github/workflows/merge-main-into-prs.yml
@@ -34,11 +34,11 @@ jobs:
g = Github(os.getenv('GITHUB_TOKEN'))
repo = g.get_repo(os.getenv('GITHUB_REPOSITORY'))
-
+
# Fetch the default branch name
default_branch_name = repo.default_branch
default_branch = repo.get_branch(default_branch_name)
-
+
for pr in repo.get_pulls(state='open', sort='created'):
try:
# Get full names for repositories and branches
diff --git a/README.md b/README.md
index ac8b63a..8872959 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@
- 
+ 
diff --git a/README.zh-CN.md b/README.zh-CN.md
index bb45872..530fed5 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -10,6 +10,7 @@
+
diff --git a/export.py b/export.py
index 5bc9238..2c697ec 100644
--- a/export.py
+++ b/export.py
@@ -91,6 +91,8 @@ MACOS = platform.system() == "Darwin" # macOS environment
class iOSModel(torch.nn.Module):
+ """An iOS-compatible wrapper for YOLOv5 models that normalizes input images based on their dimensions."""
+
def __init__(self, model, im):
"""
Initializes an iOS compatible model with normalization based on image dimensions.
diff --git a/models/common.py b/models/common.py
index 1e0ffdd..8ad53d5 100644
--- a/models/common.py
+++ b/models/common.py
@@ -71,7 +71,8 @@ def autopad(k, p=None, d=1):
class Conv(nn.Module):
- # Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation)
+ """Applies a convolution, batch normalization, and activation function to an input tensor in a neural network."""
+
default_act = nn.SiLU() # default activation
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, d=1, act=True):
@@ -91,7 +92,8 @@ class Conv(nn.Module):
class DWConv(Conv):
- # Depth-wise convolution
+ """Implements a depth-wise convolution layer with optional activation for efficient spatial filtering."""
+
def __init__(self, c1, c2, k=1, s=1, d=1, act=True):
"""Initializes a depth-wise convolution layer with optional activation; args: input channels (c1), output
channels (c2), kernel size (k), stride (s), dilation (d), and activation flag (act).
@@ -100,7 +102,8 @@ class DWConv(Conv):
class DWConvTranspose2d(nn.ConvTranspose2d):
- # Depth-wise transpose convolution
+ """A depth-wise transpose convolutional layer for upsampling in neural networks, particularly in YOLOv5 models."""
+
def __init__(self, c1, c2, k=1, s=1, p1=0, p2=0):
"""Initializes a depth-wise transpose convolutional layer for YOLOv5; args: input channels (c1), output channels
(c2), kernel size (k), stride (s), input padding (p1), output padding (p2).
@@ -109,7 +112,8 @@ class DWConvTranspose2d(nn.ConvTranspose2d):
class TransformerLayer(nn.Module):
- # Transformer layer https://arxiv.org/abs/2010.11929 (LayerNorm layers removed for better performance)
+ """Transformer layer with multihead attention and linear layers, optimized by removing LayerNorm."""
+
def __init__(self, c, num_heads):
"""
Initializes a transformer layer, sans LayerNorm for performance, with multihead attention and linear layers.
@@ -132,7 +136,8 @@ class TransformerLayer(nn.Module):
class TransformerBlock(nn.Module):
- # Vision Transformer https://arxiv.org/abs/2010.11929
+ """A Transformer block for vision tasks with convolution, position embeddings, and Transformer layers."""
+
def __init__(self, c1, c2, num_heads, num_layers):
"""Initializes a Transformer block for vision tasks, adapting dimensions if necessary and stacking specified
layers.
@@ -157,7 +162,8 @@ class TransformerBlock(nn.Module):
class Bottleneck(nn.Module):
- # Standard bottleneck
+ """A bottleneck layer with optional shortcut and group convolution for efficient feature extraction."""
+
def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):
"""Initializes a standard bottleneck layer with optional shortcut and group convolution, supporting channel
expansion.
@@ -176,7 +182,8 @@ class Bottleneck(nn.Module):
class BottleneckCSP(nn.Module):
- # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+ """CSP bottleneck layer for feature extraction with cross-stage partial connections and optional shortcuts."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
"""Initializes CSP bottleneck with optional shortcuts; args: ch_in, ch_out, number of repeats, shortcut bool,
groups, expansion.
@@ -201,7 +208,8 @@ class BottleneckCSP(nn.Module):
class CrossConv(nn.Module):
- # Cross Convolution Downsample
+ """Implements a cross convolution layer with downsampling, expansion, and optional shortcut."""
+
def __init__(self, c1, c2, k=3, s=1, g=1, e=1.0, shortcut=False):
"""
Initializes CrossConv with downsampling, expanding, and optionally shortcutting; `c1` input, `c2` output
@@ -221,7 +229,8 @@ class CrossConv(nn.Module):
class C3(nn.Module):
- # CSP Bottleneck with 3 convolutions
+ """Implements a CSP Bottleneck module with three convolutions for enhanced feature extraction in neural networks."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
"""Initializes C3 module with options for channel count, bottleneck repetition, shortcut usage, group
convolutions, and expansion.
@@ -239,7 +248,8 @@ class C3(nn.Module):
class C3x(C3):
- # C3 module with cross-convolutions
+ """Extends the C3 module with cross-convolutions for enhanced feature extraction in neural networks."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
"""Initializes C3x module with cross-convolutions, extending C3 with customizable channel dimensions, groups,
and expansion.
@@ -250,7 +260,8 @@ class C3x(C3):
class C3TR(C3):
- # C3 module with TransformerBlock()
+ """C3 module with TransformerBlock for enhanced feature extraction in object detection models."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
"""Initializes C3 module with TransformerBlock for enhanced feature extraction, accepts channel sizes, shortcut
config, group, and expansion.
@@ -261,7 +272,8 @@ class C3TR(C3):
class C3SPP(C3):
- # C3 module with SPP()
+ """Extends the C3 module with an SPP layer for enhanced spatial feature extraction and customizable channels."""
+
def __init__(self, c1, c2, k=(5, 9, 13), n=1, shortcut=True, g=1, e=0.5):
"""Initializes a C3 module with SPP layer for advanced spatial feature extraction, given channel sizes, kernel
sizes, shortcut, group, and expansion ratio.
@@ -272,7 +284,8 @@ class C3SPP(C3):
class C3Ghost(C3):
- # C3 module with GhostBottleneck()
+ """Implements a C3 module with Ghost Bottlenecks for efficient feature extraction in YOLOv5."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
"""Initializes YOLOv5's C3 module with Ghost Bottlenecks for efficient feature extraction."""
super().__init__(c1, c2, n, shortcut, g, e)
@@ -281,7 +294,8 @@ class C3Ghost(C3):
class SPP(nn.Module):
- # Spatial Pyramid Pooling (SPP) layer https://arxiv.org/abs/1406.4729
+ """Implements Spatial Pyramid Pooling (SPP) for feature extraction, ref: https://arxiv.org/abs/1406.4729."""
+
def __init__(self, c1, c2, k=(5, 9, 13)):
"""Initializes SPP layer with Spatial Pyramid Pooling, ref: https://arxiv.org/abs/1406.4729, args: c1 (input channels), c2 (output channels), k (kernel sizes)."""
super().__init__()
@@ -301,7 +315,8 @@ class SPP(nn.Module):
class SPPF(nn.Module):
- # Spatial Pyramid Pooling - Fast (SPPF) layer for YOLOv5 by Glenn Jocher
+ """Implements a fast Spatial Pyramid Pooling (SPPF) layer for efficient feature extraction in YOLOv5 models."""
+
def __init__(self, c1, c2, k=5):
"""
Initializes YOLOv5 SPPF layer with given channels and kernel size for YOLOv5 model, combining convolution and
@@ -326,7 +341,8 @@ class SPPF(nn.Module):
class Focus(nn.Module):
- # Focus wh information into c-space
+ """Focuses spatial information into channel space using slicing and convolution for efficient feature extraction."""
+
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True):
"""Initializes Focus module to concentrate width-height info into channel space with configurable convolution
parameters.
@@ -342,7 +358,8 @@ class Focus(nn.Module):
class GhostConv(nn.Module):
- # Ghost Convolution https://github.com/huawei-noah/ghostnet
+ """Implements Ghost Convolution for efficient feature extraction, see https://github.com/huawei-noah/ghostnet."""
+
def __init__(self, c1, c2, k=1, s=1, g=1, act=True):
"""Initializes GhostConv with in/out channels, kernel size, stride, groups, and activation; halves out channels
for efficiency.
@@ -359,7 +376,8 @@ class GhostConv(nn.Module):
class GhostBottleneck(nn.Module):
- # Ghost Bottleneck https://github.com/huawei-noah/ghostnet
+ """Efficient bottleneck layer using Ghost Convolutions, see https://github.com/huawei-noah/ghostnet."""
+
def __init__(self, c1, c2, k=3, s=1):
"""Initializes GhostBottleneck with ch_in `c1`, ch_out `c2`, kernel size `k`, stride `s`; see https://github.com/huawei-noah/ghostnet."""
super().__init__()
@@ -379,7 +397,8 @@ class GhostBottleneck(nn.Module):
class Contract(nn.Module):
- # Contract width-height into channels, i.e. x(1,64,80,80) to x(1,256,40,40)
+ """Contracts spatial dimensions into channel dimensions for efficient processing in neural networks."""
+
def __init__(self, gain=2):
"""Initializes a layer to contract spatial dimensions (width-height) into channels, e.g., input shape
(1,64,80,80) to (1,256,40,40).
@@ -399,7 +418,8 @@ class Contract(nn.Module):
class Expand(nn.Module):
- # Expand channels into width-height, i.e. x(1,64,80,80) to x(1,16,160,160)
+ """Expands spatial dimensions by redistributing channels, e.g., from (1,64,80,80) to (1,16,160,160)."""
+
def __init__(self, gain=2):
"""
Initializes the Expand module to increase spatial dimensions by redistributing channels, with an optional gain
@@ -422,7 +442,8 @@ class Expand(nn.Module):
class Concat(nn.Module):
- # Concatenate a list of tensors along dimension
+ """Concatenates tensors along a specified dimension for efficient tensor manipulation in neural networks."""
+
def __init__(self, dimension=1):
"""Initializes a Concat module to concatenate tensors along a specified dimension."""
super().__init__()
@@ -436,7 +457,8 @@ class Concat(nn.Module):
class DetectMultiBackend(nn.Module):
- # YOLOv5 MultiBackend class for python inference on various backends
+ """YOLOv5 MultiBackend class for inference on various backends including PyTorch, ONNX, TensorRT, and more."""
+
def __init__(self, weights="yolov5s.pt", device=torch.device("cpu"), dnn=False, data=None, fp16=False, fuse=True):
"""Initializes DetectMultiBackend with support for various inference backends, including PyTorch and ONNX."""
# PyTorch: weights = *.pt
@@ -778,7 +800,8 @@ class DetectMultiBackend(nn.Module):
class AutoShape(nn.Module):
- # YOLOv5 input-robust model wrapper for passing cv2/np/PIL/torch inputs. Includes preprocessing, inference and NMS
+ """AutoShape class for robust YOLOv5 inference with preprocessing, NMS, and support for various input formats."""
+
conf = 0.25 # NMS confidence threshold
iou = 0.45 # NMS IoU threshold
agnostic = False # NMS class-agnostic
@@ -889,7 +912,8 @@ class AutoShape(nn.Module):
class Detections:
- # YOLOv5 detections class for inference results
+ """Manages YOLOv5 detection results with methods for visualization, saving, cropping, and exporting detections."""
+
def __init__(self, ims, pred, files, times=(0, 0, 0), names=None, shape=None):
"""Initializes the YOLOv5 Detections class with image info, predictions, filenames, timing and normalization."""
super().__init__()
@@ -1047,7 +1071,8 @@ class Detections:
class Proto(nn.Module):
- # YOLOv5 mask Proto module for segmentation models
+ """YOLOv5 mask Proto module for segmentation models, performing convolutions and upsampling on input tensors."""
+
def __init__(self, c1, c_=256, c2=32):
"""Initializes YOLOv5 Proto module for segmentation with input, proto, and mask channels configuration."""
super().__init__()
@@ -1062,7 +1087,8 @@ class Proto(nn.Module):
class Classify(nn.Module):
- # YOLOv5 classification head, i.e. x(b,c1,20,20) to x(b,c2)
+ """YOLOv5 classification head with convolution, pooling, and dropout layers for channel transformation."""
+
def __init__(
self, c1, c2, k=1, s=1, p=None, g=1, dropout_p=0.0
): # ch_in, ch_out, kernel, stride, padding, groups, dropout probability
diff --git a/models/tf.py b/models/tf.py
index d79e486..59bb7e0 100644
--- a/models/tf.py
+++ b/models/tf.py
@@ -49,7 +49,8 @@ from utils.general import LOGGER, make_divisible, print_args
class TFBN(keras.layers.Layer):
- # TensorFlow BatchNormalization wrapper
+ """TensorFlow BatchNormalization wrapper for initializing with optional pretrained weights."""
+
def __init__(self, w=None):
"""Initializes a TensorFlow BatchNormalization layer with optional pretrained weights."""
super().__init__()
@@ -67,7 +68,8 @@ class TFBN(keras.layers.Layer):
class TFPad(keras.layers.Layer):
- # Pad inputs in spatial dimensions 1 and 2
+ """Pads input tensors in spatial dimensions 1 and 2 with specified integer or tuple padding values."""
+
def __init__(self, pad):
"""
Initializes a padding layer for spatial dimensions 1 and 2 with specified padding, supporting both int and tuple
@@ -87,7 +89,8 @@ class TFPad(keras.layers.Layer):
class TFConv(keras.layers.Layer):
- # Standard convolution
+ """Implements a standard convolutional layer with optional batch normalization and activation for TensorFlow."""
+
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True, w=None):
"""
Initializes a standard convolution layer with optional batch normalization and activation; supports only
@@ -118,7 +121,8 @@ class TFConv(keras.layers.Layer):
class TFDWConv(keras.layers.Layer):
- # Depthwise convolution
+ """Initializes a depthwise convolution layer with optional batch normalization and activation for TensorFlow."""
+
def __init__(self, c1, c2, k=1, s=1, p=None, act=True, w=None):
"""
Initializes a depthwise convolution layer with optional batch normalization and activation for TensorFlow
@@ -147,7 +151,8 @@ class TFDWConv(keras.layers.Layer):
class TFDWConvTranspose2d(keras.layers.Layer):
- # Depthwise ConvTranspose2d
+ """Implements a depthwise ConvTranspose2D layer for TensorFlow with specific settings."""
+
def __init__(self, c1, c2, k=1, s=1, p1=0, p2=0, w=None):
"""
Initializes depthwise ConvTranspose2D layer with specific channel, kernel, stride, and padding settings.
@@ -179,7 +184,8 @@ class TFDWConvTranspose2d(keras.layers.Layer):
class TFFocus(keras.layers.Layer):
- # Focus wh information into c-space
+ """Focuses spatial information into channel space using pixel shuffling and convolution for TensorFlow models."""
+
def __init__(self, c1, c2, k=1, s=1, p=None, g=1, act=True, w=None):
"""
Initializes TFFocus layer to focus width and height information into channel space with custom convolution
@@ -201,7 +207,8 @@ class TFFocus(keras.layers.Layer):
class TFBottleneck(keras.layers.Layer):
- # Standard bottleneck
+ """Implements a TensorFlow bottleneck layer with optional shortcut connections for efficient feature extraction."""
+
def __init__(self, c1, c2, shortcut=True, g=1, e=0.5, w=None):
"""
Initializes a standard bottleneck layer for TensorFlow models, expanding and contracting channels with optional
@@ -223,7 +230,8 @@ class TFBottleneck(keras.layers.Layer):
class TFCrossConv(keras.layers.Layer):
- # Cross Convolution
+ """Implements a cross convolutional layer with optional expansion, grouping, and shortcut for TensorFlow."""
+
def __init__(self, c1, c2, k=3, s=1, g=1, e=1.0, shortcut=False, w=None):
"""Initializes cross convolution layer with optional expansion, grouping, and shortcut addition capabilities."""
super().__init__()
@@ -238,7 +246,8 @@ class TFCrossConv(keras.layers.Layer):
class TFConv2d(keras.layers.Layer):
- # Substitution for PyTorch nn.Conv2D
+ """Implements a TensorFlow 2D convolution layer, mimicking PyTorch's nn.Conv2D for specified filters and stride."""
+
def __init__(self, c1, c2, k, s=1, g=1, bias=True, w=None):
"""Initializes a TensorFlow 2D convolution layer, mimicking PyTorch's nn.Conv2D functionality for given filter
sizes and stride.
@@ -261,7 +270,8 @@ class TFConv2d(keras.layers.Layer):
class TFBottleneckCSP(keras.layers.Layer):
- # CSP Bottleneck https://github.com/WongKinYiu/CrossStagePartialNetworks
+ """Implements a CSP bottleneck layer for TensorFlow models to enhance gradient flow and efficiency."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5, w=None):
"""
Initializes CSP bottleneck layer with specified channel sizes, count, shortcut option, groups, and expansion
@@ -289,7 +299,8 @@ class TFBottleneckCSP(keras.layers.Layer):
class TFC3(keras.layers.Layer):
- # CSP Bottleneck with 3 convolutions
+ """CSP bottleneck layer with 3 convolutions for TensorFlow, supporting optional shortcuts and group convolutions."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5, w=None):
"""
Initializes CSP Bottleneck with 3 convolutions, supporting optional shortcuts and group convolutions.
@@ -313,7 +324,8 @@ class TFC3(keras.layers.Layer):
class TFC3x(keras.layers.Layer):
- # 3 module with cross-convolutions
+ """A TensorFlow layer for enhanced feature extraction using cross-convolutions in object detection models."""
+
def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5, w=None):
"""
Initializes layer with cross-convolutions for enhanced feature extraction in object detection models.
@@ -335,7 +347,8 @@ class TFC3x(keras.layers.Layer):
class TFSPP(keras.layers.Layer):
- # Spatial pyramid pooling layer used in YOLOv3-SPP
+ """Implements spatial pyramid pooling for YOLOv3-SPP with specific channels and kernel sizes."""
+
def __init__(self, c1, c2, k=(5, 9, 13), w=None):
"""Initializes a YOLOv3-SPP layer with specific input/output channels and kernel sizes for pooling."""
super().__init__()
@@ -351,7 +364,8 @@ class TFSPP(keras.layers.Layer):
class TFSPPF(keras.layers.Layer):
- # Spatial pyramid pooling-Fast layer
+ """Implements a fast spatial pyramid pooling layer for TensorFlow with optimized feature extraction."""
+
def __init__(self, c1, c2, k=5, w=None):
"""Initializes a fast spatial pyramid pooling layer with customizable in/out channels, kernel size, and
weights.
@@ -373,7 +387,8 @@ class TFSPPF(keras.layers.Layer):
class TFDetect(keras.layers.Layer):
- # TF YOLOv5 Detect layer
+ """Implements YOLOv5 object detection layer in TensorFlow for predicting bounding boxes and class probabilities."""
+
def __init__(self, nc=80, anchors=(), ch=(), imgsz=(640, 640), w=None):
"""Initializes YOLOv5 detection layer for TensorFlow with configurable classes, anchors, channels, and image
size.
@@ -427,7 +442,8 @@ class TFDetect(keras.layers.Layer):
class TFSegment(TFDetect):
- # YOLOv5 Segment head for segmentation models
+ """YOLOv5 segmentation head for TensorFlow, combining detection and segmentation."""
+
def __init__(self, nc=80, anchors=(), nm=32, npr=256, ch=(), imgsz=(640, 640), w=None):
"""Initializes YOLOv5 Segment head with specified channel depths, anchors, and input size for segmentation
models.
@@ -450,6 +466,8 @@ class TFSegment(TFDetect):
class TFProto(keras.layers.Layer):
+ """Implements convolutional and upsampling layers for feature extraction in YOLOv5 segmentation."""
+
def __init__(self, c1, c_=256, c2=32, w=None):
"""Initializes TFProto layer with convolutional and upsampling layers for feature extraction and
transformation.
@@ -466,7 +484,8 @@ class TFProto(keras.layers.Layer):
class TFUpsample(keras.layers.Layer):
- # TF version of torch.nn.Upsample()
+ """Implements a TensorFlow upsampling layer with specified size, scale factor, and interpolation mode."""
+
def __init__(self, size, scale_factor, mode, w=None):
"""
Initializes a TensorFlow upsampling layer with specified size, scale_factor, and mode, ensuring scale_factor is
@@ -488,7 +507,8 @@ class TFUpsample(keras.layers.Layer):
class TFConcat(keras.layers.Layer):
- # TF version of torch.concat()
+ """Implements TensorFlow's version of torch.concat() for concatenating tensors along the last dimension."""
+
def __init__(self, dimension=1, w=None):
"""Initializes a TensorFlow layer for NCHW to NHWC concatenation, requiring dimension=1."""
super().__init__()
@@ -581,7 +601,8 @@ def parse_model(d, ch, model, imgsz):
class TFModel:
- # TF YOLOv5 model
+ """Implements YOLOv5 model in TensorFlow, supporting TensorFlow, Keras, and TFLite formats for object detection."""
+
def __init__(self, cfg="yolov5s.yaml", ch=3, nc=None, model=None, imgsz=(640, 640)):
"""Initializes TF YOLOv5 model with specified configuration, channels, classes, model instance, and input
size.
@@ -653,7 +674,8 @@ class TFModel:
class AgnosticNMS(keras.layers.Layer):
- # TF Agnostic NMS
+ """Performs agnostic non-maximum suppression (NMS) on detected objects using IoU and confidence thresholds."""
+
def call(self, input, topk_all, iou_thres, conf_thres):
"""Performs agnostic NMS on input tensors using given thresholds and top-K selection."""
return tf.map_fn(
diff --git a/models/yolo.py b/models/yolo.py
index d89c5da..c0dd946 100644
--- a/models/yolo.py
+++ b/models/yolo.py
@@ -70,7 +70,8 @@ except ImportError:
class Detect(nn.Module):
- # YOLOv5 Detect head for detection models
+ """YOLOv5 Detect head for processing input tensors and generating detection outputs in object detection models."""
+
stride = None # strides computed during build
dynamic = False # force grid reconstruction
export = False # export mode
@@ -127,7 +128,8 @@ class Detect(nn.Module):
class Segment(Detect):
- # YOLOv5 Segment head for segmentation models
+ """YOLOv5 Segment head for segmentation models, extending Detect with mask and prototype layers."""
+
def __init__(self, nc=80, anchors=(), nm=32, npr=256, ch=(), inplace=True):
"""Initializes YOLOv5 Segment head with options for mask count, protos, and channel adjustments."""
super().__init__(nc, anchors, ch, inplace)
@@ -214,7 +216,8 @@ class BaseModel(nn.Module):
class DetectionModel(BaseModel):
- # YOLOv5 detection model
+ """YOLOv5 detection model class for object detection tasks, supporting custom configurations and anchors."""
+
def __init__(self, cfg="yolov5s.yaml", ch=3, nc=None, anchors=None):
"""Initializes YOLOv5 model with configuration file, input channels, number of classes, and custom anchors."""
super().__init__()
@@ -332,14 +335,16 @@ Model = DetectionModel # retain YOLOv5 'Model' class for backwards compatibilit
class SegmentationModel(DetectionModel):
- # YOLOv5 segmentation model
+ """YOLOv5 segmentation model for object detection and segmentation tasks with configurable parameters."""
+
def __init__(self, cfg="yolov5s-seg.yaml", ch=3, nc=None, anchors=None):
"""Initializes a YOLOv5 segmentation model with configurable params: cfg (str) for configuration, ch (int) for channels, nc (int) for num classes, anchors (list)."""
super().__init__(cfg, ch, nc, anchors)
class ClassificationModel(BaseModel):
- # YOLOv5 classification model
+ """YOLOv5 classification model for image classification tasks, initialized with a config file or detection model."""
+
def __init__(self, cfg=None, model=None, nc=1000, cutoff=10):
"""Initializes YOLOv5 model with config file `cfg`, input channels `ch`, number of classes `nc`, and `cuttoff`
index.
diff --git a/utils/__init__.py b/utils/__init__.py
index c7ece49..185afd6 100644
--- a/utils/__init__.py
+++ b/utils/__init__.py
@@ -12,7 +12,8 @@ def emojis(str=""):
class TryExcept(contextlib.ContextDecorator):
- # YOLOv5 TryExcept class. Usage: @TryExcept() decorator or 'with TryExcept():' context manager
+ """A context manager and decorator for error handling that prints an optional message with emojis on exception."""
+
def __init__(self, msg=""):
"""Initializes TryExcept with an optional message, used as a decorator or context manager for error handling."""
self.msg = msg
diff --git a/utils/activations.py b/utils/activations.py
index 47f0a99..44bec8c 100644
--- a/utils/activations.py
+++ b/utils/activations.py
@@ -7,6 +7,8 @@ import torch.nn.functional as F
class SiLU(nn.Module):
+ """Applies the Sigmoid-weighted Linear Unit (SiLU) activation function, also known as Swish."""
+
@staticmethod
def forward(x):
"""
@@ -18,6 +20,8 @@ class SiLU(nn.Module):
class Hardswish(nn.Module):
+ """Applies the Hardswish activation function, which is efficient for mobile and embedded devices."""
+
@staticmethod
def forward(x):
"""
@@ -38,7 +42,11 @@ class Mish(nn.Module):
class MemoryEfficientMish(nn.Module):
+ """Efficiently applies the Mish activation function using custom autograd for reduced memory usage."""
+
class F(torch.autograd.Function):
+ """Implements a custom autograd function for memory-efficient Mish activation."""
+
@staticmethod
def forward(ctx, x):
"""Applies the Mish activation function, a smooth ReLU alternative, to the input tensor `x`."""
diff --git a/utils/augmentations.py b/utils/augmentations.py
index bdbe077..af4c405 100644
--- a/utils/augmentations.py
+++ b/utils/augmentations.py
@@ -18,7 +18,8 @@ IMAGENET_STD = 0.229, 0.224, 0.225 # RGB standard deviation
class Albumentations:
- # YOLOv5 Albumentations class (optional, only used if package is installed)
+ """Provides optional data augmentation for YOLOv5 using Albumentations library if installed."""
+
def __init__(self, size=640):
"""Initializes Albumentations class for optional data augmentation in YOLOv5 with specified input size."""
self.transform = None
@@ -378,7 +379,8 @@ def classify_transforms(size=224):
class LetterBox:
- # YOLOv5 LetterBox class for image preprocessing, i.e. T.Compose([LetterBox(size), ToTensor()])
+ """Resizes and pads images to specified dimensions while maintaining aspect ratio for YOLOv5 preprocessing."""
+
def __init__(self, size=(640, 640), auto=False, stride=32):
"""Initializes a LetterBox object for YOLOv5 image preprocessing with optional auto sizing and stride
adjustment.
@@ -405,7 +407,8 @@ class LetterBox:
class CenterCrop:
- # YOLOv5 CenterCrop class for image preprocessing, i.e. T.Compose([CenterCrop(size), ToTensor()])
+ """Applies center crop to an image, resizing it to the specified size while maintaining aspect ratio."""
+
def __init__(self, size=640):
"""Initializes CenterCrop for image preprocessing, accepting single int or tuple for size, defaults to 640."""
super().__init__()
@@ -424,7 +427,8 @@ class CenterCrop:
class ToTensor:
- # YOLOv5 ToTensor class for image preprocessing, i.e. T.Compose([LetterBox(size), ToTensor()])
+ """Converts BGR np.array image from HWC to RGB CHW format, normalizes to [0, 1], and supports FP16 if half=True."""
+
def __init__(self, half=False):
"""Initializes ToTensor for YOLOv5 image preprocessing, with optional half precision (half=True for FP16)."""
super().__init__()
diff --git a/utils/dataloaders.py b/utils/dataloaders.py
index 0593e52..61358eb 100644
--- a/utils/dataloaders.py
+++ b/utils/dataloaders.py
@@ -131,6 +131,8 @@ def seed_worker(worker_id):
# Inherit from DistributedSampler and override iterator
# https://github.com/pytorch/pytorch/blob/master/torch/utils/data/distributed.py
class SmartDistributedSampler(distributed.DistributedSampler):
+ """A distributed sampler ensuring deterministic shuffling and balanced data distribution across GPUs."""
+
def __iter__(self):
"""Yields indices for distributed data sampling, shuffled deterministically based on epoch and seed."""
g = torch.Generator()
@@ -259,7 +261,8 @@ class _RepeatSampler:
class LoadScreenshots:
- # YOLOv5 screenshot dataloader, i.e. `python detect.py --source "screen 0 100 100 512 256"`
+ """Loads and processes screenshots for YOLOv5 detection from specified screen regions using mss."""
+
def __init__(self, source, img_size=640, stride=32, auto=True, transforms=None):
"""
Initializes a screenshot dataloader for YOLOv5 with specified source region, image size, stride, auto, and
@@ -428,7 +431,8 @@ class LoadImages:
class LoadStreams:
- # YOLOv5 streamloader, i.e. `python detect.py --source 'rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP streams`
+ """Loads and processes video streams for YOLOv5, supporting various sources including YouTube and IP cameras."""
+
def __init__(self, sources="file.streams", img_size=640, stride=32, auto=True, transforms=None, vid_stride=1):
"""Initializes a stream loader for processing video streams with YOLOv5, supporting various sources including
YouTube.
@@ -531,7 +535,8 @@ def img2label_paths(img_paths):
class LoadImagesAndLabels(Dataset):
- # YOLOv5 train_loader/val_loader, loads images and labels for training and validation
+ """Loads images and their corresponding labels for training and validation in YOLOv5."""
+
cache_version = 0.6 # dataset labels *.cache version
rand_interp_methods = [cv2.INTER_NEAREST, cv2.INTER_LINEAR, cv2.INTER_CUBIC, cv2.INTER_AREA, cv2.INTER_LANCZOS4]
diff --git a/utils/general.py b/utils/general.py
index 57db68a..8c0b2fc 100644
--- a/utils/general.py
+++ b/utils/general.py
@@ -188,7 +188,8 @@ CONFIG_DIR = user_config_dir() # Ultralytics settings dir
class Profile(contextlib.ContextDecorator):
- # YOLOv5 Profile class. Usage: @Profile() decorator or 'with Profile():' context manager
+ """Context manager and decorator for profiling code execution time, with optional CUDA synchronization."""
+
def __init__(self, t=0.0, device: torch.device = None):
"""Initializes a profiling context for YOLOv5 with optional timing threshold and device specification."""
self.t = t
@@ -213,7 +214,8 @@ class Profile(contextlib.ContextDecorator):
class Timeout(contextlib.ContextDecorator):
- # YOLOv5 Timeout class. Usage: @Timeout(seconds) decorator or 'with Timeout(seconds):' context manager
+ """Enforces a timeout on code execution, raising TimeoutError if the specified duration is exceeded."""
+
def __init__(self, seconds, *, timeout_msg="", suppress_timeout_errors=True):
"""Initializes a timeout context/decorator with defined seconds, optional message, and error suppression."""
self.seconds = int(seconds)
@@ -239,7 +241,8 @@ class Timeout(contextlib.ContextDecorator):
class WorkingDirectory(contextlib.ContextDecorator):
- # Usage: @WorkingDirectory(dir) decorator or 'with WorkingDirectory(dir):' context manager
+ """Context manager/decorator to temporarily change the working directory within a 'with' statement or decorator."""
+
def __init__(self, new_dir):
"""Initializes a context manager/decorator to temporarily change the working directory."""
self.dir = new_dir # new dir
diff --git a/utils/loggers/__init__.py b/utils/loggers/__init__.py
index a37ae0f..92b4bcb 100644
--- a/utils/loggers/__init__.py
+++ b/utils/loggers/__init__.py
@@ -76,7 +76,8 @@ def _json_default(value):
class Loggers:
- # YOLOv5 Loggers class
+ """Initializes and manages various logging utilities for tracking YOLOv5 training and validation metrics."""
+
def __init__(self, save_dir=None, weights=None, opt=None, hyp=None, logger=None, include=LOGGERS):
"""Initializes loggers for YOLOv5 training and validation metrics, paths, and options."""
self.save_dir = save_dir
diff --git a/utils/loss.py b/utils/loss.py
index 7dc2ad3..fd5912f 100644
--- a/utils/loss.py
+++ b/utils/loss.py
@@ -14,7 +14,8 @@ def smooth_BCE(eps=0.1):
class BCEBlurWithLogitsLoss(nn.Module):
- # BCEwithLogitLoss() with reduced missing label effects.
+ """Modified BCEWithLogitsLoss to reduce missing label effects in YOLOv5 training with optional alpha smoothing."""
+
def __init__(self, alpha=0.05):
"""Initializes a modified BCEWithLogitsLoss with reduced missing label effects, taking optional alpha smoothing
parameter.
@@ -37,7 +38,8 @@ class BCEBlurWithLogitsLoss(nn.Module):
class FocalLoss(nn.Module):
- # Wraps focal loss around existing loss_fcn(), i.e. criteria = FocalLoss(nn.BCEWithLogitsLoss(), gamma=1.5)
+ """Applies focal loss to address class imbalance by modifying BCEWithLogitsLoss with gamma and alpha parameters."""
+
def __init__(self, loss_fcn, gamma=1.5, alpha=0.25):
"""Initializes FocalLoss with specified loss function, gamma, and alpha values; modifies loss reduction to
'none'.
@@ -71,7 +73,8 @@ class FocalLoss(nn.Module):
class QFocalLoss(nn.Module):
- # Wraps Quality focal loss around existing loss_fcn(), i.e. criteria = FocalLoss(nn.BCEWithLogitsLoss(), gamma=1.5)
+ """Implements Quality Focal Loss to address class imbalance by modulating loss based on prediction confidence."""
+
def __init__(self, loss_fcn, gamma=1.5, alpha=0.25):
"""Initializes Quality Focal Loss with given loss function, gamma, alpha; modifies reduction to 'none'."""
super().__init__()
@@ -101,6 +104,8 @@ class QFocalLoss(nn.Module):
class ComputeLoss:
+ """Computes the total loss for YOLOv5 model predictions, including classification, box, and objectness losses."""
+
sort_obj_iou = False
# Compute losses
diff --git a/utils/metrics.py b/utils/metrics.py
index d7a5820..85c1a70 100644
--- a/utils/metrics.py
+++ b/utils/metrics.py
@@ -122,7 +122,8 @@ def compute_ap(recall, precision):
class ConfusionMatrix:
- # Updated version of https://github.com/kaanakan/object_detection_confusion_matrix
+ """Generates and visualizes a confusion matrix for evaluating object detection classification performance."""
+
def __init__(self, nc, conf=0.25, iou_thres=0.45):
"""Initializes ConfusionMatrix with given number of classes, confidence, and IoU threshold."""
self.matrix = np.zeros((nc + 1, nc + 1))
diff --git a/utils/plots.py b/utils/plots.py
index 89ed1bb..e899ea4 100644
--- a/utils/plots.py
+++ b/utils/plots.py
@@ -29,7 +29,8 @@ matplotlib.use("Agg") # for writing to files only
class Colors:
- # Ultralytics color palette https://ultralytics.com/
+ """Provides an RGB color palette derived from Ultralytics color scheme for visualization tasks."""
+
def __init__(self):
"""
Initializes the Colors class with a palette derived from Ultralytics color scheme, converting hex codes to RGB.
diff --git a/utils/segment/dataloaders.py b/utils/segment/dataloaders.py
index c2be5f0..5f5666c 100644
--- a/utils/segment/dataloaders.py
+++ b/utils/segment/dataloaders.py
@@ -83,6 +83,8 @@ def create_dataloader(
class LoadImagesAndLabelsAndMasks(LoadImagesAndLabels): # for training/testing
+ """Loads images, labels, and segmentation masks for training and testing YOLO models with augmentation support."""
+
def __init__(
self,
path,
diff --git a/utils/segment/loss.py b/utils/segment/loss.py
index d4bc9d3..b3e76ae 100644
--- a/utils/segment/loss.py
+++ b/utils/segment/loss.py
@@ -12,7 +12,8 @@ from .general import crop_mask
class ComputeLoss:
- # Compute losses
+ """Computes the YOLOv5 model's loss components including classification, objectness, box, and mask losses."""
+
def __init__(self, model, autobalance=False, overlap=False):
"""Initializes the compute loss function for YOLOv5 models with options for autobalancing and overlap
handling.
diff --git a/utils/segment/metrics.py b/utils/segment/metrics.py
index 52ceb64..091b5b1 100644
--- a/utils/segment/metrics.py
+++ b/utils/segment/metrics.py
@@ -54,6 +54,8 @@ def ap_per_class_box_and_mask(
class Metric:
+ """Computes performance metrics like precision, recall, F1 score, and average precision for model evaluation."""
+
def __init__(self) -> None:
"""Initializes performance metric attributes for precision, recall, F1 score, average precision, and class
indices.
diff --git a/utils/torch_utils.py b/utils/torch_utils.py
index af3d046..8bf6585 100644
--- a/utils/torch_utils.py
+++ b/utils/torch_utils.py
@@ -420,7 +420,8 @@ def smart_resume(ckpt, optimizer, ema=None, weights="yolov5s.pt", epochs=300, re
class EarlyStopping:
- # YOLOv5 simple early stopper
+ """Implements early stopping to halt training when no improvement is observed for a specified number of epochs."""
+
def __init__(self, patience=30):
"""Initializes simple early stopping mechanism for YOLOv5, with adjustable patience for non-improving epochs."""
self.best_fitness = 0.0 # i.e. mAP