Computer Vision Interview
20 essential Q&A
Updated 2026
ORB
ORB: 20 Essential Q&A
FAST + oriented BRIEF—fast real-time features on CPUs and mobile.
~10 min read
20 questions
Intermediate
FASTBRIEFHammingreal-time
Quick Navigation
1
What is ORB?
⚡ easy
Answer: Oriented FAST and Rotated BRIEF—free alternative to SIFT/SURF: FAST corners, orientation from intensity centroid, steered BRIEF binary descriptor with learned pattern (rBRIEF).
2
What is FAST?
📊 medium
Answer: Compare pixel to arc of circle pixels; corner if contiguous segment brighter/darker by threshold—very fast binary tests.
3
What is BRIEF?
📊 medium
Answer: Binary string from pairwise intensity comparisons in smoothed patch—256 bits typical; match with Hamming distance.
4
ORB orientation?
📊 medium
Answer: Intensity centroid vs corner—angle of vector from keypoint to centroid gives dominant direction to steer BRIEF.
5
What is rBRIEF?
🔥 hard
Answer: Learn subset of BRIEF pairs with low correlation under rotation—better variance and discrimination than random BRIEF when oriented.
6
Scale in ORB?
⚡ easy
Answer: Image pyramid with FAST+BRIEF at each level—approximates scale invariance like other multi-scale detectors.
7
Match ORB how?
⚡ easy
Answer: Hamming distance on bitstrings—very fast with POPCNT; BFMatcher or LSH variants.
8
ORB vs BRISK?
📊 medium
Answer: BRISK uses scale-space FAST-like sampling with learned pattern; both binary; tradeoffs in pattern and scale sampling differ.
9
ORB vs SIFT?
📊 medium
Answer: ORB: faster, compact binary, less discriminative on hard wide-baseline; SIFT: float 128-D, heavier, often stronger on difficult pairs.
10
Typical ORB length?
⚡ easy
Answer: 256 bits (32 bytes)—fixed in OpenCV default; tunable via WTA_K and descriptor size params.
11
BRIEF pixel pairs?
📊 medium
Answer: Predefined or learned (x_i, y_i) locations in patch; compare I(x_i)<I(y_i) → bit—rotation steers coordinates.
12
Binary descriptor noise?
⚡ easy
Answer: Sensitive to bit flips from noise—Gaussian smoothing before sampling reduces; strong blur changes comparisons.
13
Steered BRIEF?
🔥 hard
Answer: Rotate sampling coordinates by orientation θ before comparisons—makes descriptor rotation invariant.
14
WTA in ORB OpenCV?
🔥 hard
Answer: Can use multi-point WTA to build richer binary tests—implementation detail in ORB options.
15
OpenCV?
⚡ easy
Answer:
cv2.ORB_create(nfeatures=500) → detectAndCompute.
orb = cv2.ORB_create(500)
kp, des = orb.detectAndCompute(gray, None)
16
NMS on FAST?
⚡ easy
Answer: Suppress nearby FAST responses—ORB applies score (Harris) and grid to distribute features.
17
Harris on FAST?
📊 medium
Answer: Use Harris measure on candidate FAST points to rank corner quality.
18
Why ORB on mobile?
⚡ easy
Answer: Low memory, integer/bit ops, real-time VO/SLAM on CPUs without GPU.
19
When ORB struggles?
📊 medium
Answer: Strong viewpoint change, repetitive textures, heavy motion blur—may need SIFT/AKAZE or learning methods.
20
What is AKAZE briefly?
📊 medium
Answer: Nonlinear scale space + binary descriptor—often stronger than ORB on some benchmarks, still efficient.
ORB Cheat Sheet
Detect
- FAST + pyramid
- Harris score
Describe
- rBRIEF steered
- 256 bits
Match
- Hamming
- Very fast
💡 Pro tip: ORB = FAST + orientation + steered binary BRIEF.
Full tutorial track
Go deeper with the matching tutorial chapter and code examples.