【YOLO】YOLOv8 train训练参数详解
对于我们主要关注两点batch和workers,还有个device可用可不用batch:主要对应每批次导入几个图片,影响速度,默认16
workers:线程数,默认8,对于win系统的GPU版本,必须设置为0,不然报错,cpu版本无视
device:对应cpu和gpu,默认gpu,如果未安装cuda则自动选cpu
一些解释:https://blog.csdn.net/qq_37553692/article/details/130898732
官网:https://docs.ultralytics.com/modes/train/#train-settings
ArgumentDefaultDescription
modelNoneSpecifies the model file for training. Accepts a path to either a .pt pretrained model or a .yaml configuration file. Essential for defining the model structure or initializing weights.
dataNonePath to the dataset configuration file (e.g., coco8.yaml). This file contains dataset-specific parameters, including paths to training and validation data, class names, and number of classes.
epochs100Total number of training epochs. Each epoch represents a full pass over the entire dataset. Adjusting this value can affect training duration and model performance.
timeNoneMaximum training time in hours. If set, this overrides the epochs argument, allowing training to automatically stop after the specified duration. Useful for time-constrained training scenarios.
patience100Number of epochs to wait without improvement in validation metrics before early stopping the training. Helps prevent overfitting by stopping training when performance plateaus.
batch16Batch size, with three modes: set as an integer (e.g., batch=16), auto mode for 60% GPU memory utilization (batch=-1), or auto mode with specified utilization fraction (batch=0.70).
imgsz640Target image size for training. All images are resized to this dimension before being fed into the model. Affects model accuracy and computational complexity.
saveTrueEnables saving of training checkpoints and final model weights. Useful for resuming training or model deployment.
save_period-1Frequency of saving model checkpoints, specified in epochs. A value of -1 disables this feature. Useful for saving interim models during long training sessions.
cacheFalseEnables caching of dataset images in memory (True/ram), on disk (disk), or disables it (False). Improves training speed by reducing disk I/O at the cost of increased memory usage.
deviceNoneSpecifies the computational device(s) for training: a single GPU (device=0), multiple GPUs (device=0,1), CPU (device=cpu), or MPS for Apple silicon (device=mps).
workers8Number of worker threads for data loading (per RANK if Multi-GPU training). Influences the speed of data preprocessing and feeding into the model, especially useful in multi-GPU setups.
projectNoneName of the project directory where training outputs are saved. Allows for organized storage of different experiments.
nameNoneName of the training run. Used for creating a subdirectory within the project folder, where training logs and outputs are stored.
exist_okFalseIf True, allows overwriting of an existing project/name directory. Useful for iterative experimentation without needing to manually clear previous outputs.
pretrainedTrueDetermines whether to start training from a pretrained model. Can be a boolean value or a string path to a specific model from which to load weights. Enhances training efficiency and model performance.
optimizer'auto'Choice of optimizer for training. Options include SGD, Adam, AdamW, NAdam, RAdam, RMSProp etc., or auto for automatic selection based on model configuration. Affects convergence speed and stability.
verboseFalseEnables verbose output during training, providing detailed logs and progress updates. Useful for debugging and closely monitoring the training process.
seed0Sets the random seed for training, ensuring reproducibility of results across runs with the same configurations.
deterministicTrueForces deterministic algorithm use, ensuring reproducibility but may affect performance and speed due to the restriction on non-deterministic algorithms.
single_clsFalseTreats all classes in multi-class datasets as a single class during training. Useful for binary classification tasks or when focusing on object presence rather than classification.
rectFalseEnables rectangular training, optimizing batch composition for minimal padding. Can improve efficiency and speed but may affect model accuracy.
cos_lrFalseUtilizes a cosine learning rate scheduler, adjusting the learning rate following a cosine curve over epochs. Helps in managing learning rate for better convergence.
close_mosaic10Disables mosaic data augmentation in the last N epochs to stabilize training before completion. Setting to 0 disables this feature.
resumeFalseResumes training from the last saved checkpoint. Automatically loads model weights, optimizer state, and epoch count, continuing training seamlessly.
ampTrueEnables Automatic Mixed Precision (AMP) training, reducing memory usage and possibly speeding up training with minimal impact on accuracy.
fraction1.0Specifies the fraction of the dataset to use for training. Allows for training on a subset of the full dataset, useful for experiments or when resources are limited.
profileFalseEnables profiling of ONNX and TensorRT speeds during training, useful for optimizing model deployment.
freezeNoneFreezes the first N layers of the model or specified layers by index, reducing the number of trainable parameters. Useful for fine-tuning or transfer learning.
lr00.01Initial learning rate (i.e. SGD=1E-2, Adam=1E-3) . Adjusting this value is crucial for the optimization process, influencing how rapidly model weights are updated.
lrf0.01Final learning rate as a fraction of the initial rate = (lr0 * lrf), used in conjunction with schedulers to adjust the learning rate over time.
momentum0.937Momentum factor for SGD or beta1 for Adam optimizers, influencing the incorporation of past gradients in the current update.
weight_decay0.0005L2 regularization term, penalizing large weights to prevent overfitting.
warmup_epochs3.0Number of epochs for learning rate warmup, gradually increasing the learning rate from a low value to the initial learning rate to stabilize training early on.
warmup_momentum0.8Initial momentum for warmup phase, gradually adjusting to the set momentum over the warmup period.
warmup_bias_lr0.1Learning rate for bias parameters during the warmup phase, helping stabilize model training in the initial epochs.
box7.5Weight of the box loss component in the loss function, influencing how much emphasis is placed on accurately predicting bounding box coordinates.
cls0.5Weight of the classification loss in the total loss function, affecting the importance of correct class prediction relative to other components.
dfl1.5Weight of the distribution focal loss, used in certain YOLO versions for fine-grained classification.
pose12.0Weight of the pose loss in models trained for pose estimation, influencing the emphasis on accurately predicting pose keypoints.
kobj2.0Weight of the keypoint objectness loss in pose estimation models, balancing detection confidence with pose accuracy.
label_smoothing0.0Applies label smoothing, softening hard labels to a mix of the target label and a uniform distribution over labels, can improve generalization.
nbs64Nominal batch size for normalization of loss.
overlap_maskTrueDetermines whether segmentation masks should overlap during training, applicable in instance segmentation tasks.
mask_ratio4Downsample ratio for segmentation masks, affecting the resolution of masks used during training.
dropout0.0Dropout rate for regularization in classification tasks, preventing overfitting by randomly omitting units during training.
valTrueEnables validation during training, allowing for periodic evaluation of model performance on a separate dataset.
plotsFalseGenerates and saves plots of training and validation metrics, as well as prediction examples, providing visual insights into model performance and learning progression.
6666666666666
页:
[1]