1Which component connects a Raspberry Pi camera module to the Raspberry Pi board?
Raspberry Pi camera architecture
Easy
A.CSI connector
B.Ethernet connector
C.USB connector
D.HDMI connector
Correct Answer: CSI connector
Explanation:
The Raspberry Pi camera module commonly connects through the board's CSI camera interface.
Incorrect! Try again.
2What is the main purpose of the image sensor in a camera module?
Raspberry Pi camera architecture
Easy
A.To connect to Wi-Fi networks
B.To control the display brightness
C.To capture light and form an image
D.To store video files
Correct Answer: To capture light and form an image
Explanation:
The image sensor converts incoming light into electronic image data.
Incorrect! Try again.
3Before connecting a Raspberry Pi camera ribbon cable, what should usually be done?
Setting up the Raspberry Pi camera
Easy
A.Enable Bluetooth
B.Start a video call
C.Remove the microSD card
D.Power off the Raspberry Pi
Correct Answer: Power off the Raspberry Pi
Explanation:
Powering off the board helps prevent damage while connecting the camera cable.
Incorrect! Try again.
4Which part of the Raspberry Pi camera cable must be aligned correctly with the connector?
Setting up the Raspberry Pi camera
Easy
A.The cable contacts
B.The cable outer edge
C.The cable length
D.The cable label color
Correct Answer: The cable contacts
Explanation:
The exposed contacts on the ribbon cable must face the correct direction inside the connector.
Incorrect! Try again.
5Which software feature may need to be enabled before using a Raspberry Pi camera?
Setting up the Raspberry Pi camera
Easy
A.Game controller support
B.Camera support
C.Printer support
D.Audio recording support
Correct Answer: Camera support
Explanation:
Camera support or the appropriate camera interface must be enabled for the system to access the module.
Incorrect! Try again.
6What does ISP commonly stand for in a camera system?
Camera pipeline and ISP fundamentals
Easy
A.Image Storage Program
B.Input Sensor Port
C.Image Signal Processor
D.Internet Streaming Protocol
Correct Answer: Image Signal Processor
Explanation:
An ISP processes raw sensor data into a useful image.
Incorrect! Try again.
7Which task is commonly performed by an image signal processor?
Camera pipeline and ISP fundamentals
Easy
A.White balance adjustment
B.Network cable testing
C.Keyboard layout selection
D.File compression only
Correct Answer: White balance adjustment
Explanation:
The ISP can perform operations such as white balance, noise reduction, and color processing.
Incorrect! Try again.
8What is the usual first source of data in a digital camera pipeline?
Camera pipeline and ISP fundamentals
Easy
A.Web browser
B.Keyboard
C.Image sensor
D.Speaker
Correct Answer: Image sensor
Explanation:
The image sensor first captures light and produces raw image data for later processing.
Incorrect! Try again.
9What is the purpose of converting raw camera data into a processed image?
Camera pipeline and ISP fundamentals
Easy
A.To improve image usability
B.To disconnect the camera
C.To reduce screen size
D.To disable image colors
Correct Answer: To improve image usability
Explanation:
Processing makes the captured data suitable for viewing, saving, or computer vision tasks.
Incorrect! Try again.
10What is OpenCV mainly used for?
OpenCV installation and usage
Easy
A.Computer vision tasks
B.Managing email accounts
C.Creating operating systems
D.Writing spreadsheets
Correct Answer: Computer vision tasks
Explanation:
OpenCV is a library that provides tools for image and video processing and computer vision.
Incorrect! Try again.
11Which Python package name is commonly used to install OpenCV?
OpenCV installation and usage
Easy
A.camera-tools
B.opencv-python
C.python-image-system
D.vision-linux
Correct Answer: opencv-python
Explanation:
The Python package opencv-python provides access to OpenCV functions.
Incorrect! Try again.
12Which Python statement commonly imports OpenCV?
OpenCV installation and usage
Easy
A.import cv2
B.load camera
C.using vision
D.include opencv
Correct Answer: import cv2
Explanation:
OpenCV is commonly imported in Python using the module name cv2.
Incorrect! Try again.
13Why might an image be converted to grayscale before processing?
Image preprocessing and enhancement
Easy
A.To connect the image to Wi-Fi
B.To simplify image data
C.To add more color channels
D.To increase camera voltage
Correct Answer: To simplify image data
Explanation:
A grayscale image uses intensity values instead of multiple color channels, which can simplify processing.
Incorrect! Try again.
14What does resizing an image change?
Image preprocessing and enhancement
Easy
A.Its file permissions only
B.Its network address
C.Its width and height
D.Its camera connector
Correct Answer: Its width and height
Explanation:
Resizing changes the number of pixels across the image and down the image.
Incorrect! Try again.
15Which operation can help reduce small amounts of image noise?
Image preprocessing and enhancement
Easy
A.Renaming
B.Rotating
C.Blurring
D.Printing
Correct Answer: Blurring
Explanation:
A suitable blur filter can smooth small variations and reduce noise.
Incorrect! Try again.
16What is the goal of face detection?
Face detection and recognition
Easy
A.To locate faces in an image
B.To increase image file size
C.To identify network cables
D.To change the camera lens
Correct Answer: To locate faces in an image
Explanation:
Face detection finds regions of an image that contain human faces.
Incorrect! Try again.
17How is face recognition different from face detection?
Face detection and recognition
Easy
A.Recognition only changes image size
B.Recognition identifies a person
C.Recognition removes all faces
D.Recognition controls camera power
Correct Answer: Recognition identifies a person
Explanation:
Detection finds faces, while recognition compares a detected face with known identities.
Incorrect! Try again.
18What does an object detection model usually provide for an object in an image?
Object detection using deep learning models
Easy
A.A keyboard shortcut and a password
B.A file name and a folder
C.A sound level and a voltage
D.A class and a bounding box
Correct Answer: A class and a bounding box
Explanation:
Object detectors commonly identify the object category and mark its location with a bounding box.
Incorrect! Try again.
19What is a bounding box in object detection?
Object detection using deep learning models
Easy
A.A network security rule
B.A rectangle around an object
C.A box for storing model files
D.A camera mounting bracket
Correct Answer: A rectangle around an object
Explanation:
A bounding box indicates the approximate position of a detected object in an image.
Incorrect! Try again.
20What does video streaming over a network allow a user to do?
Video streaming over a network
Easy
A.Remove all video frames
B.View video from another device
C.Replace the camera sensor
D.Increase lens brightness physically
Correct Answer: View video from another device
Explanation:
Network streaming sends video data so another connected device can view it.
Incorrect! Try again.
21Which interface normally carries image data from an official Raspberry Pi camera module to the Raspberry Pi processor?
Raspberry Pi camera architecture
Medium
A.I2C control interface
B.UART communication interface
C.SPI serial interface
D.MIPI CSI-2 interface
Correct Answer: MIPI CSI-2 interface
Explanation:
The camera sends high-speed pixel data through MIPI CSI-2. I2C is generally used to configure and control the image sensor.
Incorrect! Try again.
22A fast-moving fan appears tilted in images captured by a Raspberry Pi camera with a rolling-shutter sensor. What is the most likely cause?
Raspberry Pi camera architecture
Medium
A.Rows are exposed at different times
B.Frames are stored in reverse order
C.Pixels are transmitted through I2C
D.Colors are sampled at different depths
Correct Answer: Rows are exposed at different times
Explanation:
A rolling shutter exposes sensor rows sequentially. Motion during this interval can make rotating or moving objects appear tilted or distorted.
Incorrect! Try again.
23After connecting a camera to a recent Raspberry Pi OS installation, which command is most suitable for quickly verifying that the camera can capture a preview?
Setting up the Raspberry Pi camera
Medium
A.opencv-preview
B.rpicam-hello
C.sensor-check
D.camera-mount
Correct Answer: rpicam-hello
Explanation:
rpicam-hello opens a camera preview using the modern Raspberry Pi camera software stack and is commonly used as an initial test.
Incorrect! Try again.
24A CSI camera is not detected immediately after its ribbon cable was reconnected. What should be checked first?
Setting up the Raspberry Pi camera
Medium
A.Neural-network confidence threshold
B.OpenCV color conversion settings
C.Network router port forwarding
D.Cable orientation and connector locking
Correct Answer: Cable orientation and connector locking
Explanation:
An incorrectly oriented or loosely locked CSI ribbon cable is a common reason for camera detection failure.
Incorrect! Try again.
25What is the main purpose of the demosaicing stage in a camera image signal processor?
Camera pipeline and ISP fundamentals
Medium
A.Detecting objects from trained feature representations
B.Compressing complete frames into an H.264 stream
C.Reconstructing RGB values from filtered sensor samples
D.Transmitting raw pixels through a network socket
Correct Answer: Reconstructing RGB values from filtered sensor samples
Explanation:
A Bayer sensor records only one color component at each pixel. Demosaicing estimates the missing components to produce a full-color image.
Incorrect! Try again.
26A camera produces dark frames with little motion blur. Which adjustment is most appropriate if additional image noise is acceptable?
Camera pipeline and ISP fundamentals
Medium
A.Lower the capture resolution
B.Disable color correction
C.Increase the sensor gain
D.Reduce the exposure time
Correct Answer: Increase the sensor gain
Explanation:
Increasing gain brightens the image without requiring a longer exposure, but it also amplifies sensor noise.
Incorrect! Try again.
27Under white indoor lighting, a white object appears yellow in the captured image. Which ISP operation should primarily correct this effect?
Camera pipeline and ISP fundamentals
Medium
A.Lossless image compression
B.Temporal frame buffering
C.Automatic white balance
D.Geometric lens correction
Correct Answer: Automatic white balance
Explanation:
Automatic white balance adjusts the relative color-channel gains so that neutral objects appear neutral under different light sources.
Incorrect! Try again.
28An OpenCV frame appears blue where red is expected after being displayed with a library that expects RGB data. Which conversion should be applied?
OpenCV installation and usage
Medium
A.cv2.COLOR_BGR2RGB
B.cv2.COLOR_HSV2BGR
C.cv2.COLOR_RGB2GRAY
D.cv2.COLOR_GRAY2RGB
Correct Answer: cv2.COLOR_BGR2RGB
Explanation:
OpenCV normally stores color images in BGR order. Converting BGR to RGB corrects swapped red and blue channels in RGB-based display libraries.
Incorrect! Try again.
29A Python camera program exits, but the camera remains unavailable to another process for a short time. Which OpenCV action helps prevent this problem?
OpenCV installation and usage
Medium
A.Call cv2.cvtColor() before saving
B.Call cv2.resize() before displaying
C.Call frame.copy() after every capture
D.Call capture.release() before exiting
Correct Answer: Call capture.release() before exiting
Explanation:
The release() method closes the video-capture resource so that the camera device can be used cleanly by another process.
Incorrect! Try again.
30A program receives a valid frame from cv2.VideoCapture, but cv2.imshow() appears frozen because the event loop is not processed. Which call is needed inside the display loop?
OpenCV installation and usage
Medium
A.cv2.getBuildInformation()
B.cv2.waitKey(1)
C.cv2.destroyAllWindows()
D.cv2.imwrite(frame)
Correct Answer: cv2.waitKey(1)
Explanation:
cv2.waitKey(1) processes window events and briefly waits for keyboard input, allowing the display window to refresh.
Incorrect! Try again.
31An image of size must be resized to a width of while preserving its aspect ratio. What should the new height be?
Image preprocessing and enhancement
Medium
A. pixels
B. pixels
C. pixels
D. pixels
Correct Answer: pixels
Explanation:
The width is reduced by a factor of , so the height becomes pixels.
Incorrect! Try again.
32A grayscale image has poor contrast in some regions, while other regions already have acceptable contrast. Which technique is most suitable for local enhancement without strongly amplifying noise?
Image preprocessing and enhancement
Medium
A.CLAHE
B.Nearest-neighbor resizing
C.Global thresholding
D.Binary inversion
Correct Answer: CLAHE
Explanation:
CLAHE enhances contrast in small regions and limits histogram amplification, reducing the risk of excessive noise enhancement.
Incorrect! Try again.
33A Canny edge detector produces many false edges because the input image contains sensor noise. Which preprocessing step is most appropriate?
Image preprocessing and enhancement
Medium
A.Convert to HSV
B.Increase color saturation
C.Rotate the image
D.Apply Gaussian smoothing
Correct Answer: Apply Gaussian smoothing
Explanation:
Gaussian smoothing reduces high-frequency noise before edge detection, helping Canny avoid responding to small random intensity variations.
Incorrect! Try again.
34A face detector is missing several partially visible faces. Lowering its detection threshold is likely to produce which result?
Face detection and recognition
Medium
A.More detections and more false positives
B.Fewer detections and fewer false positives
C.More detections and faster inference
D.Fewer detections and higher resolution
Correct Answer: More detections and more false positives
Explanation:
A lower threshold accepts weaker predictions, which may recover missed faces but also increases the number of incorrect detections.
Incorrect! Try again.
35A face-recognition system represents each face as an embedding vector. How is a newly captured face commonly identified?
Face detection and recognition
Medium
A.Match its average brightness with templates
B.Compare its file size with stored images
C.Match its bounding-box area with profiles
D.Compare its embedding with stored embeddings
Correct Answer: Compare its embedding with stored embeddings
Explanation:
Recognition commonly compares the new embedding with known embeddings using a distance measure such as Euclidean or cosine distance.
Incorrect! Try again.
36A recognition system works for frontal faces but performs poorly when faces are tilted. Which preprocessing step is most likely to improve consistency?
Face detection and recognition
Medium
A.Convert all frames to binary images
B.Encode frames using H.264
C.Increase the camera frame rate
D.Align faces using detected landmarks
Correct Answer: Align faces using detected landmarks
Explanation:
Landmark-based alignment normalizes face orientation and feature positions, making recognition embeddings more consistent.
Incorrect! Try again.
37A detector produces several highly overlapping boxes for the same object. Which post-processing method should be applied?
Object detection using deep learning models
Medium
A.Non-maximum suppression
B.Background subtraction
C.Histogram equalization
D.Bilinear interpolation
Correct Answer: Non-maximum suppression
Explanation:
Non-maximum suppression keeps the strongest box and removes lower-confidence boxes that overlap it beyond an IoU threshold.
Incorrect! Try again.
38An object-detection model expects a square input, but the camera frame is widescreen. Which preprocessing method best preserves object shapes?
Object detection using deep learning models
Medium
A.Convert the frame to grayscale
B.Stretch directly to the square size
C.Resize proportionally and add padding
D.Crop equal strips from all sides
Correct Answer: Resize proportionally and add padding
Explanation:
Proportional resizing followed by padding, often called letterboxing, preserves the aspect ratio and avoids geometric distortion.
Incorrect! Try again.
39A Raspberry Pi object detector is accurate but too slow for the required frame rate. Which change is most likely to improve speed while retaining the same general task?
Object detection using deep learning models
Medium
A.Disable all confidence filtering
B.Increase the model input resolution
C.Save every frame as PNG
D.Use a quantized lightweight model
Correct Answer: Use a quantized lightweight model
Explanation:
A lightweight quantized model requires fewer computations and less memory, usually improving inference speed on resource-constrained hardware.
Incorrect! Try again.
40An MJPEG stream sends JPEG frames per second, and each frame averages KB. Approximately what application-level bandwidth is required?
Video streaming over a network
Medium
A. Mbps
B. Mbps
C. Mbps
D. Mbps
Correct Answer: Mbps
Explanation:
The stream sends KB/s. Multiplying by gives approximately Kb/s, or Mbps, excluding protocol overhead.
Incorrect! Try again.
41A Raspberry Pi camera is connected through the CSI interface, while image processing is performed largely by the SoC multimedia subsystem. Which consequence is most important when diagnosing a high-resolution capture failure?
Raspberry Pi camera architecture
Hard
A.The GPU can increase the physical resolution of the attached sensor
B.The sensor supplies pixels, but memory bandwidth and ISP throughput can limit usable modes
C.The CSI connector determines the sensor's maximum exposure time
D.The camera automatically bypasses the ISP during still-image capture
Correct Answer: The sensor supplies pixels, but memory bandwidth and ISP throughput can limit usable modes
Explanation:
The sensor defines available pixel modes, but the CSI receiver, memory bandwidth, and ISP processing capacity determine whether those modes can be captured and processed reliably.
Incorrect! Try again.
42A rolling-shutter Raspberry Pi camera observes a rapidly rotating fan. Which artifact is most likely, and which architectural property explains it?
Raspberry Pi camera architecture
Hard
A.Uniform darkening, caused by ISP color correction
B.Color banding, caused by Bayer interpolation
C.Radial blur, caused by JPEG quantization
D.Geometric skew, caused by sequential row exposure
Correct Answer: Geometric skew, caused by sequential row exposure
Explanation:
A rolling shutter exposes different image rows at different times, so fast motion changes the apparent geometry across the frame.
Incorrect! Try again.
43A camera application must provide a low-latency preview and periodically save high-quality still images. Which design best uses the camera architecture?
Raspberry Pi camera architecture
Hard
A.Switch the sensor off between preview frames to reduce processing load
B.Process the preview only after each still image has finished encoding
C.Capture every frame as a full-resolution JPEG and resize for preview
D.Use a low-resolution video stream for preview and a still stream for photographs
Correct Answer: Use a low-resolution video stream for preview and a still stream for photographs
Explanation:
Separate stream configurations allow the preview path to remain responsive while the still path uses higher resolution and quality when required.
Incorrect! Try again.
44After physically connecting a camera, the preview command reports that no camera is detected. Which diagnostic sequence is most appropriate?
Setting up the Raspberry Pi camera
Hard
A.Increase JPEG quality, then reboot the camera process
B.Change the image format to PNG, then disable automatic exposure
C.Verify ribbon orientation and seating, inspect detection output, then test a supported mode
D.Install OpenCV first, then replace the power supply
Correct Answer: Verify ribbon orientation and seating, inspect detection output, then test a supported mode
Explanation:
Detection failures are commonly caused by cable orientation, connector seating, or unsupported software configuration; these should be checked before application-level changes.
Incorrect! Try again.
45A camera works at 640x480 but fails at its advertised maximum resolution. Which explanation is most technically defensible?
Setting up the Raspberry Pi camera
Hard
A.The selected mode may exceed available bandwidth, memory, or pipeline constraints
B.The operating system always limits CSI cameras to VGA
C.The lens focal length prevents large image buffers
D.The camera sensor cannot expose pixels larger than VGA
Correct Answer: The selected mode may exceed available bandwidth, memory, or pipeline constraints
Explanation:
Advertised sensor resolution does not guarantee that every complete pipeline configuration can sustain the required throughput and buffer allocation.
Incorrect! Try again.
46A manually focused camera produces sharp images at close range but blurred images for distant objects. Which adjustment is most appropriate?
Setting up the Raspberry Pi camera
Hard
A.Increase the frame rate to force a shorter focal distance
B.Increase JPEG compression to sharpen edge transitions
C.Disable the ISP so the lens receives more exposure data
D.Rotate the lens focus mechanism toward the distance-focused position
Correct Answer: Rotate the lens focus mechanism toward the distance-focused position
Explanation:
Focus is controlled optically by lens position. Image encoding and ISP settings cannot compensate for an incorrectly focused lens across object distances.
Incorrect! Try again.
47Why can a raw Bayer frame appear strongly tinted when opened directly as if it were an ordinary RGB image?
Camera pipeline and ISP fundamentals
Hard
A.The CSI link reverses the red and blue channels
B.The sensor has already applied an incorrect gamma curve
C.Each pixel stores only one color sample under a Bayer color filter pattern
D.JPEG decoding is required before any sensor color exists
Correct Answer: Each pixel stores only one color sample under a Bayer color filter pattern
Explanation:
Raw Bayer data contains a mosaic of single-channel samples. Demosaicing reconstructs RGB pixels, followed by operations such as white balance and color correction.
Incorrect! Try again.
48An image has correct brightness but a strong blue cast under warm indoor lighting. Which ISP operation primarily addresses this problem?
Camera pipeline and ISP fundamentals
Hard
A.H.264 entropy coding
B.Temporal denoising
C.White-balance estimation and channel gain adjustment
D.Lens shading correction
Correct Answer: White-balance estimation and channel gain adjustment
Explanation:
White balance estimates the scene illuminant and scales color channels so neutral objects appear neutral under different lighting conditions.
Incorrect! Try again.
49A scene contains a bright window and a dark indoor subject. Why might enabling HDR or multi-exposure processing improve the subject while introducing motion artifacts?
Camera pipeline and ISP fundamentals
Hard
A.It combines exposures with different dynamic ranges, which can misalign moving content
B.It replaces the lens with a wider aperture, which causes object duplication
C.It lowers sensor resolution by interpolating every exposure independently
D.It removes all saturated pixels by converting them to grayscale
Correct Answer: It combines exposures with different dynamic ranges, which can misalign moving content
Explanation:
HDR methods combine differently exposed images. Motion between exposures can create ghosting or misregistration unless motion compensation is effective.
Incorrect! Try again.
50An OpenCV application reports that the camera backend opens successfully but returns empty frames intermittently. Which first change best isolates the cause?
OpenCV installation and usage
Hard
A.Increase the neural-network confidence threshold
B.Convert every frame to grayscale before checking its validity
C.Check the return status and frame contents before any processing operation
D.Replace all NumPy arrays with Python lists
Correct Answer: Check the return status and frame contents before any processing operation
Explanation:
Capture failure must be distinguished from processing failure. Validity checks prevent downstream operations from masking the actual camera or backend problem.
Incorrect! Try again.
51A script works in a system Python shell but fails inside a virtual environment with an import error for OpenCV. What is the most likely cause?
OpenCV installation and usage
Hard
A.OpenCV requires a separate HDMI driver for each environment
B.The virtual environment is using a different interpreter or lacks the OpenCV package
C.The image sensor has changed its Bayer pattern
D.The camera ribbon cable is reversed
Correct Answer: The virtual environment is using a different interpreter or lacks the OpenCV package
Explanation:
Python packages are interpreter- and environment-specific. Confirming the active executable and installing a compatible OpenCV package in that environment resolves this class of error.
Incorrect! Try again.
52A BGR frame from OpenCV is passed directly to a model trained with RGB images. The model's confidence decreases substantially. Which correction is required?
OpenCV installation and usage
Hard
A.Rotate the image by 90 degrees before inference
B.Apply JPEG compression before inference
C.Swap the blue and red channels before normalization
D.Convert the image to a larger integer type only
Correct Answer: Swap the blue and red channels before normalization
Explanation:
OpenCV commonly represents color images as BGR, while many trained models expect RGB. Channel order must match the training preprocessing pipeline.
Incorrect! Try again.
53A grayscale image has poor global contrast because most pixels occupy a narrow intensity interval, but illumination varies across the scene. Which method is generally most suitable?
Image preprocessing and enhancement
Hard
A.Nearest-neighbor enlargement followed by sharpening
B.Histogram equalization applied independently to every video frame
C.CLAHE with controlled tile contrast limits
D.Global thresholding with a fixed midpoint
Correct Answer: CLAHE with controlled tile contrast limits
Explanation:
Contrast-limited adaptive histogram equalization improves local contrast while limiting noise amplification, making it more suitable than a single global mapping under uneven illumination.
Incorrect! Try again.
54A face detector misses small faces after an image is resized for speed. Which change most directly addresses the recall problem while controlling computational cost?
Image preprocessing and enhancement
Hard
A.Lower JPEG quality to create more edge pixels
B.Use a larger detector input or image pyramid only when needed
C.Apply a stronger median blur before detection
D.Remove normalization so dark pixels remain unchanged
Correct Answer: Use a larger detector input or image pyramid only when needed
Explanation:
Small objects lose representational detail during downscaling. Increasing detector resolution or selectively processing higher-resolution regions improves recall at an explicit compute cost.
Incorrect! Try again.
55An edge detector produces many false edges in a low-light image. Which preprocessing choice is most likely to reduce false responses without eliminating all object boundaries?
Image preprocessing and enhancement
Hard
A.Apply denoising before gradient calculation
B.Replace all pixels below the mean with white
C.Increase saturation before converting to grayscale
D.Use a larger JPEG compression ratio before filtering
Correct Answer: Apply denoising before gradient calculation
Explanation:
Noise creates rapid intensity changes that resemble edges. Moderate denoising before gradient computation can suppress these responses while preserving larger structural boundaries.
Incorrect! Try again.
56A recognition system performs well on enrollment images but confuses people under different lighting and head poses. Which redesign best separates detection from recognition concerns?
Face detection and recognition
Hard
A.Align detected faces, extract embeddings, and compare them with a calibrated threshold
B.Train the system using only blurred face thumbnails
C.Increase the detector's bounding-box size until identities become unique
D.Use a face detector alone and treat every detected box as an identity
Correct Answer: Align detected faces, extract embeddings, and compare them with a calibrated threshold
Explanation:
Detection locates faces, while recognition compares identity representations. Alignment and embedding-based comparison improve robustness to pose and illumination changes.
Incorrect! Try again.
57A face-recognition system must reject unknown people, but its nearest-neighbor classifier always assigns the closest enrolled identity. What change is necessary?
Face detection and recognition
Hard
A.Sort enrolled identities alphabetically before comparison
B.Disable face alignment to preserve more background context
C.Use a distance or similarity threshold with an explicit unknown outcome
D.Increase the number of image color channels from three to four
Correct Answer: Use a distance or similarity threshold with an explicit unknown outcome
Explanation:
Nearest-neighbor matching always returns a class. A calibrated threshold is required to reject matches whose similarity is insufficient.
Incorrect! Try again.
58A detector produces multiple overlapping boxes for one object. Which post-processing operation is intended to retain the strongest box while suppressing redundant detections?
Object detection using deep learning models
Hard
A.Bilinear interpolation
B.Non-maximum suppression
C.Histogram stretching
D.Batch normalization
Correct Answer: Non-maximum suppression
Explanation:
Non-maximum suppression ranks boxes by confidence and removes boxes whose overlap with a stronger box exceeds a chosen IoU threshold.
Incorrect! Try again.
59A model trained on 640x640 letterboxed images receives a 1280x720 frame. Which preprocessing error is most likely to corrupt predicted box coordinates?
Object detection using deep learning models
Hard
A.Applying non-maximum suppression after confidence filtering
B.Using RGB values instead of grayscale values
C.Ignoring the scale and padding offsets when mapping boxes back
D.Keeping the original frame rate during inference
Correct Answer: Ignoring the scale and padding offsets when mapping boxes back
Explanation:
Letterboxing changes coordinates through scaling and padding. Predictions must be inversely transformed to align with the original frame.
Incorrect! Try again.
60A Raspberry Pi streams video over Wi-Fi. Viewers observe increasing delay even though the displayed video remains smooth. Which change most directly limits latency growth?
Video streaming over a network
Hard
A.Use bounded queues and discard stale frames when processing falls behind
B.Increase the stream resolution while preserving the same bitrate
C.Use an unbounded receive buffer to avoid dropped frames
D.Encode every frame as a lossless image before transmission
Correct Answer: Use bounded queues and discard stale frames when processing falls behind
Explanation:
Unbounded buffering preserves old frames but causes latency to grow. Bounded queues maintain freshness by dropping frames that are no longer useful.
Incorrect! Try again.
Did this save you a night before the exam?
LPU Notes is free, and it stays free. Ads cover part of the server bill.
The rest comes out of a student's own pocket: the domain, the storage,
and keeping the site up through the weeks everyone needs it at once.
The payment button didn't load. An ad blocker or a filtered network is the usual reason.
to try again.
Nothing here is ever locked, and nothing unlocks. Chip in only if it was worth it.
What it pays for →