gst-inference 做出错误的推断

如何解决gst-inference 做出错误的推断

我在基于 arm64 的 ubuntu18.04 板上安装了 gststreamer+gstinference+r2inference+tensorflow-lite。 一切顺利，直到我开始使用 gst-launch cmd 推断图像。

---

IMAGE_FILE=/home/pi/cat.jpg
LABELS='labels.txt'
MODEL_LOCATION='graph_inceptionv2.tflite'
GST_DEBUG=inceptionv2:6 gst-launch-1.0 \
multifilesrc location=$IMAGE_FILE start-index=0 stop-index=0 loop=false  ! jpegparse ! jpegdec ! videoconvert ! videoscale ! videorate ! queue ! jpegenc ! filesink location=capture1.jpeg

Which gave out a result of:
---------------------------------------
Setting pipeline to PAUSED ...
0:00:00.135856000 21061   0x55a93450a0 INFO             inceptionv2 gstinceptionv2.c:293:gst_inceptionv2_start:<net> Starting Inception v2
Pipeline is prerollED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
0:00:00.173726875 21061   0x55a92faf70 LOG              inceptionv2 gstinceptionv2.c:217:gst_inceptionv2_preprocess:<net> Preprocess
0:00:00.743937542 21061   0x55a92faf70 LOG              inceptionv2 gstinceptionv2.c:229:gst_inceptionv2_postprocess_old:<net> Postprocess
0:00:00.744053917 21061   0x55a92faf70 LOG              inceptionv2 gstinferencedebug.c:74:gst_inference_print_highest_probability:<net> Highest probability is label 106 : (0.002603)
0:00:00.744067626 21061   0x55a92faf70 LOG              inceptionv2 gstinceptionv2.c:254:gst_inceptionv2_postprocess_new:<net> Postprocess Meta
0:00:00.744423459 21061   0x55a92faf70 LOG              inceptionv2 gstinferencedebug.c:111:gst_inference_print_predictions: 
{
  id : 0,enabled : True,bBox : {
    x : 0
    y : 0
    width : 224
    height : 224
  },classes : [
    {
      Id : 0
      Class : 106
      Label : koala,koala bear,kangaroo bear,native bear,Phascolarctos cinereus
      Probability : 0.002603
      Classes : 250
    },],predictions : [
    
  ]
}

--------------------------------------
Said it was koala bear,But this is a 224*224 egytian cat images which was correctly inferenced by R2inference sample program.

/home/sfq/test/inception -i /home/pi/cat.jpg -m ./graph_inceptionv2.tflite -s 224
Loading Model: ./graph_inceptionv2.tflite
Setting model to engine
Loading image: /home/pi/cat.jpg
Configuring frame
Starting engine
Predicting...
Highest probability is label 286 (0.791615)
Stopping engine

我的 Gst-inference 和 R2Inference 配置之间有什么问题？？请帮忙。