Recap Object Detection

Which of the following is NOT a commonly used object detection algorithm?

• YOLO

• SSD

• R-CNN

• ResNet

What does the "anchor box" represent in object detection models like YOLO and SSD?

• A ground truth label

• A proposed region for classification

• A pre-defined box used to predict bounding boxes

• A type of data augmentation

What is the purpose of the Intersection over Union (IoU) metric?

• To improve model training time

• To resize input images correctly

• To measure overlap between two boxes

• To detect the object class name

Which of the following best describes the YOLO detection approach?

• Detects objects in multiple stages

• Uses a grid to predict bounding boxes

• Uses selective region proposals

• Predicts bounding boxes with a decoder

In Faster R-CNN, what does the RPN do?

• It classifies object categories

• It generates object region proposals

• It filters out noisy bounding boxes

• It calculates total model loss

What is the function of Non-Maximum Suppression (NMS)?

• Combines object feature vectors

• Eliminates duplicate box predictions

• Reduces training dataset size

• Converts masks to bounding boxes

Which metric is most commonly used to evaluate object detection models?

• Accuracy over all images

• Average classification score

• Mean Average Precision (mAP)

• Mean Square Bounding Error

What is a major difference between classification and object detection?

• Detection classifies without labels

• Detection uses deeper neural nets

• Detection finds object locations too

• Detection avoids feature extraction