FlowVis

0.1 verified Sat May 09 auth: no python

A Python package for easy visualization of optical flow fields, providing utilities to convert flow vectors into color-coded images for quick inspection. Current version 0.1, released as an initial alpha with simple generation and display functions; maintenance is sporadic.

pip install flow-vis

Common errors

error ImportError: cannot import name 'flow_viz' from 'flow_viz' ↓

cause The correct import is from the module 'flow_vis', not 'flow_viz'. The function is named 'flow_viz' inside 'flow_vis'.

fix

Use: from flow_vis import flow_viz

error ModuleNotFoundError: No module named 'cv2' ↓

cause OpenCV is required but not installed automatically.

fix

Install opencv-python: pip install opencv-python

error ValueError: operands could not be broadcast together with shapes ↓

cause The flow array must have shape (H, W, 2). If you provide a scalar or wrong dimensions, it fails.

fix

Ensure your flow array is a 3D numpy array with last dimension size 2.

Warnings

gotcha The input flow array must be of shape (H, W, 2) with the last dimension containing (dx, dy). The values are not normalized; the function will clip to a default range [-20, 20] pixels. ↓

fix Normalize your flow values to a sensible range (e.g., clip to [-20,20]) or adjust the function's internal parameters.

deprecated The package uses 'cv2' (OpenCV) internally but does not declare it as a hard dependency, causing import errors if OpenCV is missing. ↓

fix Install opencv-python separately: pip install opencv-python.

gotcha The function 'flow_viz' returns an RGB image in uint8 format with values in [0,255]. Do not expect normalized float output. ↓

fix Use the resulting image directly with OpenCV or matplotlib; for float display, convert to float32 and divide by 255.

Imports

flow_viz
wrong
```
import flow_viz
```
correct
```
from flow_vis import flow_viz
```
flow_viz is a function inside the flow_vis module, not a top-level module.
make_colorwheel
```
from flow_vis import make_colorwheel
```

Quickstart

Generate a color-coded visualization from a flow array.

import numpy as np
from flow_vis import flow_viz

# Create a dummy flow field (height, width, 2)
flow = np.random.uniform(-10, 10, (100, 200, 2)).astype(np.float32)

# Visualize
rgb = flow_viz(flow)

# Optionally display using OpenCV
import cv2
cv2.imshow('Optical Flow', rgb)
cv2.waitKey(0)
cv2.destroyAllWindows()