Machine Learning with OpenCV
Machine Learning is a branch of Artificial Intelligence and concerned with the question how to make machines able to learn from data. The core idea is to enable a machine to make intelligent decisions and predictions based on experiences from the past. Algorithms of Machine Learning require interdisciplinary knowledge and often intersect with topics of statistics, mathematics, physics, pattern recognition and more.
OpenCV (Open Source Computer Vision) is a library for computer vision and comes with a machine learning library for:
- Decision Trees
- Boosting
- Support Vector Machines
- Expectation Maximization
- Neural Networks
- …
Finding simple examples to get started is difficult, so I wrote a document and a program for the C++ Machine Learning API of OpenCV. You can download it from my github account at: https://github.com/bytefish/opencv/.
Experiments
| Parameter | Value |
|---|---|
| Trainingdata size | 200 |
| Testdata size | 2000 |
y = 2x
| Predictor | Accuracy |
|---|---|
| Support Vector Machine | 0.99 |
| Multi Layer Perceptron (2, 10, 15, 1) | 0.994 |
| k-Nearest-Neighbor (k = 3) | 0.9825 |
| Normal Bayes | 0.9425 |
| Decision Tree | 0.923 |
Plot
y = sin(10x)
| Predictor | Accuracy |
|---|---|
| Support Vector Machine | 0.913 |
| Multi Layer Perceptron (2, 10, 15, 1) | 0.6855 |
| k-Nearest-Neighbor (k = 3) | 0.9 |
| Normal Bayes | 0.632 |
| Decision Tree | 0.886 |
Plot
y = tan(10x)
| Predictor | Accuracy |
|---|---|
| Support Vector Machine | 0.7815 |
| Multi Layer Perceptron (2, 10, 15, 1) | 0.5115 |
| k-Nearest-Neighbor (k = 3) | 0.8195 |
| Normal Bayes | 0.542 |
| Decision Tree | 0.9155 |
Plot
