Find the source of the labeling error when detection is correct
Question: A vision system correctly draws a bounding box around a dog, but it labels a Labrador as a beagle. If the detector localized the animal properly, which component is responsible for the mistake?
Sample answer: The mistake comes from the breed classifier. Since the detector found the animal correctly, the wrong breed label must be caused by the part of the system that assigns the breed.
Key points:
- The breed classifier is the faulty component.
- The detector is not at fault because it produced the correct bounding box.
Rubric: The answer must identify the breed classifier as the source of the error and explain that the detector is not responsible because it localized the animal correctly.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI
Related
A wildlife app first locates an animal in a photo and then predicts its breed. If the locator returns the right bounding box but the final breed is wrong for a Persian cat, which stage caused the mistake?
True or False: If a route planner selects the wrong route even though its road-quality estimator ranks the roads correctly, the estimator should be blamed.
If a flower detector draws the correct bounding box but the system still names the wrong flower species, the error should be attributed to the flower _____ classifier.
Attributing a wrong species label after a correct detection
True or False: If the localization step finds the correct region and the system still mislabels the animal, the final classifier is at fault.
If the upstream crop is correct, a wrong parcel-condition label should be attributed to the _____.
Match each pipeline situation to the correct source of the error in a bird recognition system.
Put the troubleshooting steps in order for tracing a warehouse label mistake.
A pet image pipeline gives the wrong breed label. The detector draws a tight box around the animal. Where should debugging concentrate?
True or False: If an object detector finds the cat with the right bounding box, it should still be blamed for a later breed mistake.
Downstream Error Attribution in a Two-Stage Cat Classifier
Match each pipeline outcome to the correct error-attribution conclusion.
Order the reasoning used to decide whether the license-plate reader or the character classifier caused the mistake.
Explain why a correct sign box makes the final labeling error the classifier's responsibility.
Attributing a correct detection with a wrong subclass label
Find the source of the labeling error when detection is correct