New Custom Compontent - Image Processing - Object Detection - DOODS

snowzach · December 3, 2019, 4:25am

Hey, everything looks pretty good. Maybe the still image is not coming through okay. Maybe lower the confidence really low to see if it gets a hit on anything.

snowzach · December 3, 2019, 4:25am

Try raising your confidence. It’s not perfect by any means. You could also try another model.

firstof9 · December 3, 2019, 12:27pm

Even higher it was missing the people, like it was getting the wrong image some how.

snowzach · December 3, 2019, 2:07pm

So depending on the detector, it resizes to 300x300. If the aspect is very wide, maybe it’s distorting the image too much. Eventually I want to make the add-on crop the global detection area. I haven’t had time to work on it in a bit. Life getting in the way of my hobbies. lol

firstof9 · December 3, 2019, 3:13pm

It’s like it’s returning the correct detection boxes, but the wrong image behind them, if that makes sense at all.

BlazeYaSmartHome · December 5, 2019, 4:58pm

I think its just detecting stuff it thinks is people, the confidence doesn’t look very high. (40%) etc This happened a lot with mine till i switched to tesnorflow and upped the confidence quite a bit.

snowzach · December 5, 2019, 5:46pm

Yah, I agree with @BlazeYaSmartHome, tensorflow lite mobilenet SSD vs say inception on tensorflow is night and day difference in accuracy

firstof9 · December 6, 2019, 12:53am

I’m using the default detector:

  detectors:
    - name: default
      type: tflite
      modelFile: models/coco_ssd_mobilenet_v1_1.0_quant.tflite
      labelFile: models/coco_labels0.txt
      numThreads: 4
      numConcurrent: 4

I’m game for suggestions
Thank you.

pollinolas · December 17, 2019, 3:06am

hello, I bought a google coral usb, I’m still waiting to receive it, right now I’m using the hassio addon without acceleration, when I want to use google coral with the addon, would it just change the line “hwAccel” to true? Or would I have to do something else? Regards, and thanks for your work

snowzach · December 17, 2019, 3:26am

You need to use a model that’s designed for the edge tpu. You can download them here: https://coral.ai/models/

pollinolas · December 17, 2019, 3:29am

Thanks for answer so quickly, where I should copy this models?
Which model will better for person recognized?

snowzach · December 17, 2019, 4:11am

Are you using docker or hass.io?

pollinolas · December 17, 2019, 4:22am

I’m using hassio on docker

pollinolas · December 18, 2019, 3:54am

Can I put the new models in the directory that I want or has to be in the default one? Because in my system there is no doods folder inside the opt or share folder
For example:
/config/www/models/xxx.tflite
Or
/local/models/xxx.tflite

snowzach · December 18, 2019, 1:33pm

Hey @pollinolas check out the docs for the hassio addon here: https://github.com/snowzach/hassio-addons/tree/master/doods

You can put them in /share/doods and there are configuration examples on that page.

danbutter · December 20, 2019, 12:17am

May I ask how you are doing this? I have one of the same cams and a few others that are also dahua.
I actually have BI, but the pullpoint subscription isn’t working for them…only my hikvision cams.
So if you have a way to go straight from the cam to HA that would be great.

Thanks

Juggler · December 20, 2019, 4:08pm

I’m using this python program that logs Dahua camera events to MQTT:

github.com

johnnyletrois/dahua-watch/blob/4d7b3e03119ebf8c72cf1673785ec368ace2ec39/watch.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import logging
import socket
import pycurl
import time
import shlex
import subprocess
import paho.mqtt.client as mqtt

ALARM_DELAY = 30

#PLAY_TEMPLATE = "gst-launch-1.0 playbin uri=\"rtsp://{user}:{pass}@{host}:554/cam/realmonitor?channel=1&subtype=2\" latency=300000000 audio-sink=\"autoaudiosink sync=false\""
URL_TEMPLATE = "http://{host}:{port}/cgi-bin/eventManager.cgi?action=attach&codes=%5B{events}%5D"

CAMERAS = [
	{
		"host": "CAM_1_IP",
		"port": 80,
		"user": "USERNAME",

This file has been truncated. show original

danbutter · December 21, 2019, 10:16pm

Thanks for the lead. I followed that link and a couple mins later ended up at the amcrest component which works for dahua cams. Well two types of mine that I tried anyway.
For anyone else interested it is here:

danbutter · December 22, 2019, 12:54am

Okay I’m no programmer, but I’m trying to learn here. I just started today and got it working using the hassio (.103.2) addon with all default settings. Yay!
Thing is my wife’s beloved hydrangea shows up as a person in one cam and a car in the other. Same plant too! Just different angles from two different cams. Overall not very accurate.

So how do I replace a model? Looking at the install docs I see:

{
      "name": "inception",
      "type": "tensorflow",
      "modelFile": "/share/doods/faster_rcnn_inception_v2_coco_2018_01_28.pb",
      "labelFile": "/share/doods/coco_labels1.txt",
      "numThreads": 1,
      "numConcurrent": 1,
      "hwAccel": false
    }

yet when I download one from here (I’ve tried a few…maybe I’m getting wrong kind)
https://github.com/tensorflow/models/blob/3635527dc66cdfe7270e5b3086858db7307df8a3/research/object_detection/g3doc/detection_model_zoo.md
they don’t have the same file with it after extracting it. There is a .pb, but the name is different and no labels file.
I’ve even tried the one linked by the OP in post 104: New Custom Compontent - Image Processing - Object Detection - DOODS
Any suggestions on what I’m doing wrong?
Thanks

pollinolas · December 23, 2019, 8:05pm

hello, I already have it working with google coral, in my case the directory was in :
./var/lib/docker/overlay2/ce6cdb5a533fd7b056e98cd6e2b004ffae2d249e7bc92c7aef17c18a91bdacdb/diff/opt/doods/models/
I had to search the .tflite file to find the estrange directory