How to use multiple types of input in deepstream?

My model has 2 kinds, image and camera inside and outside parameters. How does this work in deepstream?

Can you elaborate more details about your model? How many input layers and how many output layers? What is the input for each input layer? What is the output layers?

Do you mean your model inputs are some images and camera parameters? What kind of camara parameters? Lens, Retinal Plane,…?

The network is 3D object detection, the input of the network is an image and the internal and external parameters of this image. The internal parameters are the internal parameters of the camera, and the external parameters refer to the conversion and translation matrix of the camera to the radar coordinate system. Internal and external parameters are used to transform network features.

Do you mean you only need one image inferencing? What kind of image(2D or 3D)? Will the camera parameters change with time?

yes,Yes. My model is what you say it is。

You may consider implement the TensorRT inferencing by yourself and then integrate the inferencing with nvdsvideotemplate since you need so many customization features.


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.