More info about NvDCF

Hi, where can I find more information about NvDCF? I read about it in the plugin manual for DeepStream 4 but it only mentions “discriminative filtering based” algorithm. Are there any additional resources where I can read about it? I couldn’t find any paper or code for that as well.

1 Like

May I know what kind of information about DCF you want to find?

I mean general literature, like how does it work, what is the algorithm, and possibly source code so I can decide if I should be using some other machine learning based tracker (like DeepSORT, or FairMOT etc.)

This is the info about algorithm our three low level tracker used, you also can get from plugin manual, as for the tracker algorithm principle, you can google search it. there many infomation about.

The three reference low level libraries support different algorithms:

•The KLT tracker uses a CPU-based implementation of the Kanade Lucas Tomasi (KLT) tracker algorithm. This library requires no configuration file.

•The Intersection of Union (IOU) tracker uses the intersection of the detector’s bounding boxes across frames to determine the object’s unique ID. This library takes an optional configuration file.

•The Nv-adapted Discriminative Correlation Filter (NvDCF) tracker uses a correlation filter-based online discriminative learning algorithm, coupled with a Hungarian algorithm for data association in multi-object tracking. This library accepts an optional configuration file.

source code not open source.