Computer Vision DY0-001 Practice Questions (Page 1)

Questions 1–5

1. 1expert · hard

   A company needs to extract text from images of receipts. The receipts have a consistent layout but are printed on thermal paper, which can fade and have varying contrast. The text is printed in a standard font. The team wants to extract the total amount, which is always in a fixed location. Which OCR approach is most robust to the fading and contrast issues?

   • APreprocess the image with contrast enhancement and use a template-matching OCR for the total amount
   • BUse a neural-network OCR without preprocessing
   • CUse a statistical OCR without preprocessing
   • DUse template matching without preprocessing
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   Which output is typically produced by an object detection model?

   • AA single class label for the entire image
   • BA pixel-level mask for each object
   • CA trajectory path for each object
   • DA list of object classes with bounding box coordinates
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which of the following best describes the primary purpose of Optical Character Recognition (OCR) technology?

   • ATo identify and classify objects within an image
   • BTo compress image files for storage efficiency
   • CTo convert scanned or photographed text images into machine-readable text
   • DTo enhance the visual quality of images for human viewing
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   What is the primary purpose of object tracking in computer vision?

   • ATo convert text in images into machine-readable text
   • BTo estimate the trajectory of an object across video frames
   • CTo segment an image into regions
   • DTo classify objects in a single image
   Reveal Solution Discussion Select an answer first
6. 5expert · hard

   A company is building a system to detect potholes from a camera mounted on a city bus. The bus travels at varying speeds, and the camera captures images at different angles and lighting conditions. The team has a small labeled dataset of potholes. They need a model that runs in real time on the bus's onboard computer. Which approach is most effective?

   • ATrain a Faster R-CNN model from scratch with no augmentation
   • BUse a sliding-window classifier with a deep CNN
   • CUse a color-based segmentation approach to find potholes
   • DFine-tune a pretrained YOLO model with augmentation for brightness, rotation, and scale
   Reveal Solution Discussion Select an answer first