Recognizer Types
Object Class Recognition supports several types of recognizer. All use neural networks but have different characteristics.
The default recognizer is the Generation1
recognizer. This was the sole type of recognizer in Media Server versions 12.3 and earlier. This type of recognizer is the fastest to train but the other recognizers provide equal or better accuracy and are much faster at run-time. Choose Generation1
only if you need to train the recognizer on a CPU.
The training time for Generation2
, Generation3
, and Generation4
recognizers is impractical on a CPU, so Micro Focus recommends that you train the following types of recognizer on a machine with a supported GPU. You can use a trained recognizer with or without a GPU but Micro Focus recommends using a GPU for best performance. If you run recognition on a GPU, Generation2
, Generation3
, and Generation4
recognizers are all much faster than a Generation1
recognizer, which allows you to set much shorter sample intervals.
Generation2
recognizers offer much better accuracy than aGeneration1
recognizer.Generation2
recognizers were introduced in Media Server 12.4 and Micro Focus now recommends usingGeneration4
recognizers instead.Generation3
recognizers are the fastest at run-time, but accuracy is slightly lower than aGeneration2
orGeneration4
recognizer.Generation4
recognizers offer the best accuracy, especially for recognizers that you train yourself. TheGeneration4
recognizer type was introduced in Media Server 12.9.