Join Our 5-Week ML/AI Engineer Interview Bootcamp 🚀 led by ML Tech Leads at FAANGs
Back to Questions
239. Image normalization
easy
Generalsenior
Image normalization is a simple but important preprocessing step in computer vision that makes pixel values comparable across images. In this problem, you’ll implement per-channel normalization for an RGB image represented as a NumPy array.
The normalization formula for each channel is:
Requirements
Implement the function
python
Rules:
- Normalize each channel using the given mean and std.
- Treat
meanandstdas length-3 arrays corresponding to[R, G, B]. - Use NumPy for computation.
- Do not use any prebuilt normalization utilities (e.g.,
torchvision.transforms.Normalize). - Keep the original shape
(H, W, 3).
Example
python
Output:
python
Input Signature
| Argument | Type |
|---|---|
| std | np.ndarray |
| mean | np.ndarray |
| image | np.ndarray |
Output Signature
| Return Name | Type |
|---|---|
| value | np.ndarray |
Constraints
Use NumPy.
Keep output shape exactly (H, W, 3).
No prebuilt normalization utilities.
Hint 1
Use NumPy broadcasting: (H, W, 3) array minus (3,) array works automatically.
Hint 2
result = (image - mean) / std is all you need if shapes align correcty (last dim matches).
Hint 3
No loops are needed.
Roles
ML Engineer
AI Engineer
Companies
GeneralLevels
senior
entry
Tags
numpy-broadcasting
image-preprocessing
array-shapes
normalization
33 people are solving this problem
Python EditorLn 1, Col 1