HEF model input problem

General
1

Hello I’m having trouble running fast_depth on hailo15h.

I used the following pipeline to make sure that the inputs are going to the model properly.

PIPELINE="gst-launch-1.0 \
    hailofrontendbinsrc config-file-path=$frontend_config_file_path ! \
    queue leaky=no max-size-buffers=$max_buffers_size ! \
    video/x-raw,format=NV12,width=224,height=224 ! \
    videoconvert !  identity silent=false ! \
    video/x-raw,format=RGB,width=224,height=224 ! \
    hailonet hef-path=$hef_path vdevice-group-id=device0 \
    ${additional_parameters}"

eval $PIPELINE

The models are as follows.

root@hailo15:~/depth_estimation# hailortcli parse-hef ./resources/fast_depth.hef 
Architecture HEF was compiled for: HAILO15H
Network group name: fast_depth, Single Context
    Network name: fast_depth/fast_depth
        VStream infos:
            Input  fast_depth/input_layer1 UINT8, NHWC(224x224x3)
            Output fast_depth/conv20 UINT16, NHWC(224x224x1)

I’m running fast_depth.hef in the model Zoo as above, but it seems that the input is not going in the model properly.

Did I miss or do anything wrong?

If there is a file you need, I will upload it as much as possible.