Pre-Trained Object Detectors

HPE may provide pre-trained object detectors that you can use with Media Server to detect objects belonging to certain generic categories in images and videos.

The following detectors are currently available from the Big Data Download Center, in the package PretrainedModels_MediaServer_<VERSION>.zip. When you download this package, ensure that <VERSION> matches the version of Media Server that you are using.

File name Description
ObjectDetector_HeadAndShoulder.dat An object detector for detecting and counting people. This detector differs from ObjectDetector_Person.dat because it has been trained to detect only the head and shoulder region, which is useful when you want to detect and count people in a crowded area.
ObjectDetector_Person.dat An object detector for detecting people. Multiple people can be detected in an image at a time.
ObjectDetector_RoadScene.dat An object detector that contains training for detecting instances of cars, vans and people in images and videos. Single or multiple instances of each of these objects can be located in an image at a time.

For information about how to import a detector into your training database, see Import a Detector.