What are the latest approaches for object detection and which (if applicable) are the machine learning algorithms used (i.e. SVM, Adaboost, Neural Networks)?
well in detection per-se you just named them essentially. More recently however combined approaches such as detection per parts, symbolic segmentation or so have been emerging, but the principal techniques AFAIK remained the same
Considering all of efficacy, code availability and research trends, you might consider:
For more complex scenarios (e.g. deformable objects) - The following paper, as a starting point and more recent follow-up work.
Felzenszwalb, Pedro F., Ross B. Girshick, David McAllester, and Deva Ramanan. "Object detection with discriminatively trained part-based models." Pattern Analysis and Machine Intelligence, IEEE Transactions on 32, no. 9 (2010): 1627-1645.
For scenarios more heavily constrained in the nature of the recognition scenario:
The following paper (and more recent derivative work):
Viola, Paul, and Michael J. Jones. "Robust real-time face detection." International journal of computer vision 57, no. 2 (2004): 137-154.
Also if you're feeling ambitious, models that employ deep neural nets are showing impressive results and are especially in vogue at the moment.
See: http://deeplearning.net/
HOG+SVM, Haar+Boosting classifier are more successful methods. However, you can also try part based detection/recognition.
SVM, Adaboost, or Neural Networks form the machine learning classifier stage. Before that, a suitable feature vector must be created via HOG, Haar etc. HOG + SVM or Haar + boosting are simply the pioneers, but HOG + boosting has been proven to be faster, and patented.
- Badges
- Science topic