Possibility of processing a bundle of yolov3 + video detector based on NCS2

[Karmeo]

Hello!
I want to run a script on NCS 2 based on yolov3 detector on tensorflow framework and video tracker (Kalman, Deepsort, LSTM, CAMShift)
Does your instrument support such bundles?

[Zulkifli_Intel]

Hello Ann Tolmacheva,

Greetings to you.

In order for you to utilize the NCS 2, you need to optimize your model to Intermediate Representation (IR) format using the OpenVINO Toolkit.

Are you trying to run 2 different models on one NCS 2? Please elaborate more on your query to help us understand it better.

Regards,

Zulkifli

[Karmeo]

I am going to implement the calculation of the trajectory of motion of objects on your platform. As a detection, I use a trained yolo3 model. I learned how to convert it to work in OpenVINO. Now I would like to add a motion tracker (Kalman filter, Dipsort (Kalman filter + distance calculation), CAMshift, LSTM). The first three methods do not confuse me, since these are methods from openCV, but I am also interested in using LSTM. These are already layers of the recurrent model.

[Vladimir_Dudnik · Accepted solution]

@Karmeo you also might be interested to look at existing OpenVINO Open Model Zoo demo applications, which implements tracking, although they may use different algorithms for implementation.

Multi camera multi target Python demo

Pedestrian tracker C++ demo

Regards,
  Vladimir

[Zulkifli_Intel]

Hello Ann Tolmacheva,

This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question.

Regards,

Zulkifli