How to install open cv step by step
guide_Bidyut Maishal.
Here’s a clear, practical, step-by-step guide to install OpenCV on Windows (works for Python
3.x). I’ll cover the easiest pip method, the conda alternative, verify steps, common
troubleshooting, and short notes on advanced (build-from-source / CUDA) if you need GPU
support.
1) Prerequisites
Python installed (3.7+) and working. Verify in a new terminal:
python --version
Use a virtual environment for projects (strongly recommended).
2) Create & activate a virtual environment (recommended)
Open Command Prompt or PowerShell and run:
# create project folder
mkdir my_opencv_project
cd my_opencv_project
# create venv
python -m venv venv
# activate (PowerShell)
.\venv\Scripts\Activate.ps1
# or activate (cmd.exe)
venv\Scripts\activate
When active your prompt shows (venv).
Upgrade pip, setuptools, wheel:
python -m pip install --upgrade pip setuptools wheel
3) Install OpenCV (simple & quick — pip)
There are two common pip packages:
opencv-python — core OpenCV (most users).
opencv-contrib-python — OpenCV + extra (“contrib”) modules (SIFT, xfeatures2d, etc).
Install one of these (don’t install both at the same time — uninstall one first if switching):
# core OpenCV
pip install opencv-python
# OR, for contrib modules (recommended if you need extra features)
pip install opencv-contrib-python
For headless servers (no GUI):
pip install opencv-python-headless
# or
pip install opencv-contrib-python-headless
4) Verify the installation
Run Python and try importing OpenCV and printing the version:
python - <<'PY'
import cv2
print("OpenCV version:", cv2.__version__)
PY
Or start an interactive Python shell:
>>> import cv2
>>> cv2.__version__
'4.x.x'
If that prints a version, installation worked.
5) Quick runtime tests (image + webcam)
Save a test image test.jpg in your folder and run:
# show_image.py
import cv2
img = cv2.imread('test.jpg') # load image
if img is None:
raise SystemExit("Failed to load test.jpg")
cv2.imshow('Test', img)
cv2.waitKey(0)
cv2.destroyAllWindows()
Run:
python show_image.py
Webcam capture test:
# webcam_test.py
import cv2
cap = cv2.VideoCapture(0) # 0 = default camera
if not cap.isOpened():
raise SystemExit("Cannot open camera")
ret, frame = cap.read()
if not ret:
raise SystemExit("Failed to grab frame")
cv2.imshow('Frame', frame)
cv2.waitKey(0)
cap.release()
cv2.destroyAllWindows()
Notes:
cv2.imshow opens a GUI window — this works on normal Windows desktop. It won’t work
inside some remote terminals or WSL without an X server.
If cap.isOpened() is False, check camera permissions/drivers or change index 0→1.
6) Optional: Install additional helpful packages
pip install numpy matplotlib pillow
(OpenCV uses NumPy; many tutorials use matplotlib for plots.)
7) Conda alternative (recommended if you use Anaconda/Miniconda)
If you prefer conda (often simpler for compiled libs):
conda create -n cvenv python=3.10
conda activate cvenv
conda install -c conda-forge opencv
Conda-forge builds often handle dependencies and codecs better.
8) Video codecs & ffmpeg (if video read/write fails)
If cv2.VideoCapture or cv2.VideoWriter has problems with some formats, install ffmpeg and
add it to PATH:
1. Download ffmpeg static build for Windows (from ffmpeg.org builds).
2. Unzip and add the bin folder (contains ffmpeg.exe) to your Windows PATH.
3. Restart your terminal.
9) Troubleshooting (common errors & fixes)
ImportError: DLL load failed or module 'cv2' has no attribute 'imshow':
Install Microsoft Visual C++ Redistributable (x64). Download from Microsoft (search
"Visual C++ Redistributable for Visual Studio"). Install the x64 version matching your
Python bitness.
Ensure Python (32-bit vs 64-bit) matches the pip wheel; use 64-bit Python for
modern wheels.
Multiple OpenCV packages conflict:
Uninstall first, then reinstall:
pip uninstall opencv-python opencv-contrib-python opencv-python-headless -y
pip install opencv-contrib-python
Camera not opening:
Check camera drivers, privacy settings (Windows → Settings → Privacy → Camera), try
different index (0,1).
cv2.imshow doesn't open window in WSL:
WSL doesn't provide Windows GUI by default. Use Windows Python or configure an X
server/GUI for WSL or use headless display (saving images to disk).
Video codecs missing:
Install ffmpeg or use conda-forge build.
10) Advanced: Build OpenCV from source (when you need CUDA/GPU support or custom
options)
If you need GPU (CUDA) support or custom compile flags, you must build from source. High-
level steps:
1. Install prerequisites: Visual Studio (with Desktop dev. workload), CMake, NVIDIA CUDA
toolkit (if GPU), Python dev headers, Git.
2. Clone OpenCV & opencv_contrib:
3. git clone https://github.com/opencv/opencv.git
4. git clone https://github.com/opencv/opencv_contrib.git
5. Create a build folder, run CMake GUI or CLI, set options:
PYTHON_EXECUTABLE → your python.exe in venv
OPENCV_EXTRA_MODULES_PATH → path to opencv_contrib/modules
Enable WITH_CUDA (if you have CUDA and compatible GPU)
6. Configure, generate Visual Studio solution, then build (Release config).
7. After build, copy generated cv2.pyd to your venv Lib\site-packages or install via
generated wheel.
This process is non-trivial — ask me if you want a step-by-step build script for your exact
CUDA + Visual Studio versions.
11) Recommended workflow summary
1. Create venv → activate.
2. pip install opencv-contrib-python numpy matplotlib (or use conda).
3. Verify import cv2 and cv2.__version__.
4. Install ffmpeg if you work with many video formats.
5. If you need GPU acceleration, plan to build from source.
Want me to:
Give you an exact copy-paste PowerShell script that creates a venv and installs OpenCV
+ test files, or
Help you build OpenCV with CUDA for your GPU (tell me your GPU model, CUDA
version, Visual Studio version), or
Provide a small sample project (README + code) you can clone and run?
