public class IAnomalibSegModel : IModelSpecialized model for detecting and segmenting anomalies in images. 专门用于检测和分割图像中异常的模型。
Processes model outputs to generate anomaly masks and confidence scores. 处理模型输出以生成异常掩码和置信度分数。
| IAnomalibSegModel | Initializes a new instance of the anomaly segmentation model 初始化异常分割模型的新实例 |
| ModelInferenceProfiler |
Performance profiler recording timing metrics
记录时间指标的性能分析器
(继承自 IModel。) |
| Dispose |
Releases model resources
释放模型资源
(继承自 IModel。) |
| Equals | Determines whether the specified object is equal to the current object. (继承自 Object。) |
| Finalize | Allows an object to try to free resources and perform other cleanup operations before it is reclaimed by garbage collection. (继承自 Object。) |
| GetHashCode | Serves as the default hash function. (继承自 Object。) |
| GetType | Gets the Type of the current instance. (继承自 Object。) |
| Lerp | Optimized linear interpolation between two values 两个值之间的优化线性插值 |
| MemberwiseClone | Creates a shallow copy of the current Object. (继承自 Object。) |
| Postprocess |
Processes raw model outputs into structured anomaly segmentation results
将原始模型输出处理为结构化的异常分割结果
(重写 IModelPostprocess(DataTensor, ImageAdjustmentParam)) |
| Predict(ListObject) |
Performs batch prediction on multiple inputs
对多个输入执行批量预测
(继承自 IModel。) |
| Predict(Object) | Predicts anomalies in input image and returns segmentation results 预测输入图像中的异常并返回分割结果 |
| PredictAsync |
Asynchronously performs prediction on single input
异步执行单个输入预测
(继承自 IModel。) |
| Preprocess |
Preprocesses input image into model-compatible tensor format
将输入图像预处理为模型兼容的张量格式
(重写 IModelPreprocess(Object, ImageAdjustmentParam)) |
| Sigmoid | Fast approximation of sigmoid function (3x faster than Math library) Sigmoid函数的快速近似计算（比Math库快3倍） |
| ToString | Returns a string that represents the current object. (继承自 Object。) |
| config |
Model configuration parameters
模型配置参数
(继承自 IModel。) |
| engine |
Inference engine instance
推理引擎实例
(继承自 IModel。) |
| predictorTimer |
Timer measuring different prediction phases
测量不同预测阶段的计时器
(继承自 IModel。) |