Precision
The Precision is the percentage of detected features or classes that are correct, or in other words, match the labeled feature A feature is a visually distinguishable area in an image. Features typically represent something of interest for the application (a defect, an object, a particular component of an object). or class.
Precision represents the accuracy of positive predictions. It calculates how often the tool can correctly predict the positive values. Precision is calculated as the number of true positive predictions divided by the sum of true positive and false positive predictions.
- A neural network with a low Precision score typically fails to correctly identify the features that should have been detected from the test data, and so it returns many false positive results.
- A neural network with a high Precision score typically succeeds to correctly identify the features from given the test data, but if combined with low Recall, there can be many false negative predictions.
The ideal statistical results for typical inspection cases includes high Precision and high Recall scores.
Precision Calculation of Blue Locate and Blue Read
For the Blue Locate and Blue Read tools, Precision is the percentage of detected features that are not in the training Training is the process that your tool, which is a neural network, is learning about the features (pixels) based on the labels you made. For example, a tool will learn the defect/normal pixels in each image based on the defect/normal labels you drew. The goal of the tool Training is learning enough to give the correct inspection results of whether an unseen image is defective or not. The key to training is to ensure that you include all possible variations within your training set, and that your images are accurately labeled. Training times vary by the application, tool setup and the GPU in the PC being used to train the network. set and that correctly match the labeled features. For example, a Precision score of 90% for character “A” means that the tool correctly identifies "A" against all other character features with a 90% chance given the test data.
|
|
|
|
Labeled Image |
Precision Results |
For the Blue Read tool, using a model A specific spatial arrangement of a set of features (Blue Locate and Blue Read tools only.) During a post-processing step, the Blue Locate and Blue Read tools can fit all of the features detected in an image to the models defined for the tool. The overall pose and identity of the model is then returned. can help filter out spurious matches. When using a model, a low Precision score can be less problematic.
|
|
|
|
No model |
Using a model |
Precision Calculation of Green Classify
For the Green Classify tool, Precision is calculated as the percentage of detected classes that match the labeled class.
The tool calculates Precision as the percentage of correctly classified views predicted as class i:
For example, a Precision of 90% for the class i means that the neural network will confuse an image of class i with a different class 1 out of 10 times.