Running Advance demos

The same Python and C++ demo applications can be used to run multiple inference models and also work with multiple inputs with just simple changes in the config file.

From a repo of input sources, output sources and models one can define advance dataflows which connect them in various configurations. Details on configuration file parameters can be found in Demo Configuration file

Single input multi inference demo

Here is an example of a single-input, multi-inference demo which takes a camera input and run multiple networks on each of them.

root@tda4vm-sk:/opt/edge_ai_apps/apps_python# ./app_edgeai.py ../configs/single_input_multi_infer.yaml

Sample output for single input, multi inference demo is as shown below,

Fig. 12 Sample output showing single input, mutli-inference output

We can specify the output window location and sizes as shown in the configuration file,

flows:
    flow0: [input0,model0,output0,[160,90,800,450]]
    flow1: [input0,model1,output0,[960,90,800,450]]
    flow2: [input0,model2,output0,[160,540,800,450]]
    flow3: [input0,model3,output0,[960,540,800,450]]

Multi input multi inference demo

Here is an example of a multi-input, multi-inference demo which takes a camera input and video input and runs multiple networks on each of them.

root@tda4vm-sk:/opt/edge_ai_apps/apps_python# ./app_edgeai.py ../configs/multi_input_multi_infer.yaml

Sample output for multi input, multi inference demo is as shown below,

Fig. 13 Sample output showing multi-input, mutli-inference output

We can specify the output window location and sizes as shown in the configuration file,

flows:
    flow0: [input0,model1,output0,[160,90,800,450]]
    flow1: [input0,model2,output0,[960,90,800,450]]
    flow2: [input1,model0,output0,[160,540,800,450]]
    flow3: [input1,model3,output0,[960,540,800,450]]

We can also share the same model instance to do same inference on differnt inputs. This will reduce the memory foot prints when compred to creating multiple instace of same model. Below is an example for the same,

flows:
    flow0: [input0,model1,output0,[160,90,800,450]]
    flow1: [input0,model2,output0,[960,90,800,450]]
    flow2: [input1,model1,output0,[160,540,800,450]]
    flow3: [input1,model2,output0,[960,540,800,450]]

Here model1 and model2 is shared across.