FREE License Plate Reading, Face Recognition, and Object Detection for Blue Iris - Full Walkthrough
HTML-код
- Опубликовано: 6 авг 2024
- Add free ALPR (automated license plate recognition), facial recognition, and general computer vision object detection to Blue Iris using the free and open source CodeProject AI.
Blue Iris Software: bit.ly/3dwlGRg
Blue Iris $10 off coupon to make the price $69.95 (not sure how long it will be good for): PERS-10BUCKS
ALPR Zoom Camera
- AmazonUS: amzn.to/3owYeLz
- AmazonCA: amzn.to/4a6oEX4
Reolink Doorbell
- AmazonUS: amzn.to/41Iv7mT
- AmazonCA: amzn.to/3V7wifG
- AmazonUK: amzn.to/49I1mqZ
Hikvision Panoramic Camera
- AmazonUS: amzn.to/3UVbHcc
- AmazonCA: amzn.to/3TrPmnz
Blue Iris Hardware Suggestion ($450 Total):
Dell 7040
- AmazonUS: amzn.to/3mQ6CWi
- AmazonCA: amzn.to/3Vspcmn
WD Purple Surveillance HDD
- AmazonUS: amzn.to/3LdQ046
- AmazonCA: amzn.to/4cawfFW
- AmazonUK: amzn.to/43bv6Kr
SATA Power Splitter
- AmazonUS:: amzn.to/3Licoco
- AmazonCA: amzn.to/3Tcuu2m
- AmazonUK: amzn.to/3PcXFRD
GTX1650 low profile
- AmazonUS: amzn.to/3UPwuOB
Website Write Up: www.thesmarthomehookup.com/fr...
Dahua Commands:
Get current focus status: {camera ip}/cgi-bin/devVideoInput.cgi?action=getFocusStatus
Set Zoom/Focus via URL: {camera ip}/cgi-bin/devVideoInput.cgi?action=adjustFocus&focus={focus level}&zoom={zoom level}
Set Day Mode via URL: {camera ip}: /cgi-bin/configManager.cgi?action=setConfig&VideoInMode[0].Config[0]=0
Set Night Mode via URL: {camera ip}: /cgi-bin/configManager.cgi?action=setConfig&VideoInMode[0].Config[0]=1
MQTT JSON:
{"plate":"&PLATE", "AlertImgPath":"&ALERT_PATH", "Alert_AI":"&MEMO", "Date":"%Y-%m-%d %H:%M:%S","Camera":"&CAM"}
*As an Amazon Affiliate I earn a commission on qualifying purchases at no cost to you*
Follow me on Twitter: / thehookup1
Join me on Facebook: / 473812443269387
Support my channel directly:
Patreon: / thehookup
Music by www.BenSound.com
00:00 Introduction
00:42 Blue Iris Software and Hardware
01:55 Blue Iris and CodeProject Installation
02:29 Blue Iris General Settings
05:45 CodeProjectAI General Setup
07:22 Adding ONVIF Cameras to Blue Iris
09:29 Motion Detection Settings
10:56 Computer Vision Settings
12:28 Blue Iris Recording Options
13:36 Motion Troubleshooting
15:06 Facial Recognition Setup
16:53 License Plate Recognition Setup
24:29 Legal Disclaimer 
Наука
As someone who has used blue iris for almost a decade, this is by far one of the most comprehensive videos showcasing the power of this software. Well done.
I would love to see a comparison video(series) between the (affordable) self-hosted NVR options. Like Blue Iris, Frigate, Kerebos, ZoneMinder, Viseron, Milestone, ...
I specifically searched for the latest blue iris video to write this. I feel that frigate has come a long way and it deserves another look.
Oh man I've been waiting for this video. Your past Blue Iris videos helped me setup my instance and it's been working great. I've been itching for some upgrades to it though and this video will help me do that. Thank you! I can only handle so many projector videos lol
The holy grail of Blue Iris setup videos in 2023. Excellent stuff man.
Just way to fast !
25 minutes packed with great information! Thank you for a very detailed, thorough, well explained and helpful guide.
Perfect timing. Awesome video. Thanks dude for quality content with no time wasting fluff.
I’ve been an avid BlueIris user for the past 7 years and have probably spent 200+ hours tinkering with settings over that time (mostly for fun, trial & error on a 16 camera setup) and this video by far is the best tutorial I’ve seen to date. Well presented, easy to understand and although I’m not looking for license plate recognition you provided several tips on how to speed up AI recognition and face recognition that already have me tinkering with new settings like a kid in a candy shop. Anyways, A+ on the video, subscribed and enjoy your content….
I watched your 2 year-old video and things were going okay but noted you had a newer one and moved to your 1 year-old video two days ago. Things have been working great, but I'm 100% going to see what changes I can make because while it's working great I think it can do a lot better. Thank you for all your help!
Attention all users who are looking for the Artificial Intelligence Button and cannot find it under the trigger tab! Please note that the button has been relocated to the alert tab.
Thanks. UI redesign is the bane of RUclips tutorials.
@@TheHookUp Yeah it can be frustrating.
THANKS...I was second guessing my sanity.
WOW, that is a crazy amount of features for $70. And a very great job of covering ALL the parameters !
Still the best walk through 8 months later. Helpful. Concise. Comprehensible.
the amount of questions i had and challenges were all solved by this beautiful video.. masterful work. Now i can focus on getting my night time working as intended.
Great Vid! Love the pace of your videos. The content is so dense so there's no waiting to get to information I need. I need it all!. I find myself constantly rewinding and rewatching which, I'm sure, is good for your video stats. Nice work.
As Always, Rob never disappoints and Always Hits the High Marks ! Thanks again for the comprehensive guide.
This was a long one!
@@TheHookUp ... and a great one!
What a master class. I used it for updating my Blue Iris configuration I set up about a year ago, and I added Face Recognition.
This is great timing was just about to update to another computer that uses Blue Iris and AI tools after your last video years back and which matches to the specs you stated here look forward to making the changes talked about here. Thanks again for the great videos as always.
I’ve had Blue IRIS FOR YEARS. I love it… but not being experienced security camera tech I fumbled through it. Now seeing your video I can expand my experience on settling up some of the many options Blue Iris gives you. Thanks for the great video!!!
perfect timing, saving this video to my home automation playlist, thanks for all your hard work!
Thank you!! I've been waiting for an updated video from you about this! Keep up the great work!
Perfect timing on this video, I was just thinking how I wish you would make an update to that old video, especially with the new AI. Thank you
Much appreciated. Just started migrating my blue iris to a proxmox server and this is a very nicely timed update for that. Thank you
Amazing video. This channel always has such mindblowingly good content.
hands down the best tutorial ever made on youtube about this topic! +100 rep for you!
Now we're back into the good stuff. Great video!
Excellent work. Helped tweak and optimise a lot of my setup. Would love to see a future video look into the push notification capabilities of BlueIris
Thank you for the great video! Security company said they could sell me a computer with Blue Iris already installed and ready to go. Though it sounded like it was going to cost thousands. Now I feel confident I can do it myself.
Excellent video Rob! Well done! 👏
I always LOVE these videos!!!! Give us ALL the Blue Iris stuffs!!!!!!
This is awesome and well presented information. I have set up a dahua LPR camera and couldnt quite figure out how to get the night/day switch to work. Thank you so much!
This is EXACTLY what I was looking for.
Thank you!
Thanks for the awesome video! I fixed a lot of problems I was having and learned a ton just from this one video 👍
It’s incredible that all of this can be done locally
Crazy world we live in.
Locally-ish - the CodeProjectAI software still needs to talk to its home IP on port 443 to be able to analyze images, especially faces, or it won't process. So far, though, that's the only IP I've had to open to my Blue Iris machine which, unless I'm updating, has no general internet access whatsoever. Aside from that one IP on port 443, the entire thing is locally processed and accessed.
@@joevining2603 This is highly concerning. Why is no one talking about this? That's a massive security and privacy risk.
You giving video access of your daily life over to AI, and then letting that AI talk with outside sources.
Its home IP meaning the local machine where it is installed? That’s still local.
@@TheHookUp I've tried several times to reply to this providing the public IP information to where it reaches out, but the comments keep getting deleted.
Great informative update to this NVR software. Thanks a bunch for making it.
Amazing video! so much work put into this video. Thanks :D
Great video. I used your other BI setup video to get started and used the Project AI that came bundled with BI instead of using Deepstack right out of the box. Biggest issue I ran into with Project AI was updating it was a bit of a pain. Download the update. Then uninstall the current version. Install and start the update. Then you need to go in and download and install the modules. It would not give me the option to install YOLO v5.6.2 GPU (CUDA) until after I downloaded and installed YOLO v5.3.1. Then after v5.6.2 was installed and running you need to go back in and stop v5.3.1. Functionality was kaput until I figured this out just this afternoon.
Keep the videos coming!
So glad you dropped this video, neighbors have been wanting alpr.
I'd like to see you do a roadside cam with additional IR
You mean run ethernet out to the road? No doubt that would produce great results, but might be pushing the boundaries of what my neighbors find acceptable :)
@@TheHookUp or as close to the road as possible, down low. Like at the base of one your trees.
Or dont make it permanent, just as a demo of the best case scenario for ALPR and compare it to the results you are currently getting.
So beyond what I feel I'm willing to do, so I'll have to stick with watching the whole video and giving a thumbs up.
Start small!
another excellent video sir. Such a huge treasure trove of info.
Great tutorial. Love that you have a write up.
I found this to be very helpful. Thank you for sharing 👍
Amazing video !
Next video, use this for automation in home assistant :p
Keep it up ! Thank you so much for this !
Excellent info/video nice clear and precise got my ALPR working from your video thanks..
I currently have an older Amazon Blink wireless system with 5 cameras. Since installation, I haven't been bothered even once by any of the local scam artists that used to bother me. That is fantastic in and of itself but the system has lots of limitations (I know, you get what you pay for!) Thank you so much for this detailed explanation of how I can set up a better system that I can have more control over. Everything breaks eventually. I'd like to work on building a new better system before I eventually replace Blink.
Awesome video man, what a good amount of information!
Good stuff and very well said! Thanks for making this video. 😀
This is simply a brilliant video!
Would love to see a continuation or part 2 of this where you dive more into the AI and the alerting side more and how to do notifications.
I can't say it enough.. Thank you for this video.
on point, what i needed!
Awesome video!
Yes! Thanks, Rob.
very thorough! thank you
Great job!
Okay, I was bitchy last time cause you seemed to focus on non- privacy based solutions.. But it seems you have more then redeemed yourself... Thank you from the bottom of my heart for focusing on local storage and privacy minded hardware.. Great job, I'm glad I subscribed please do more like this...😊
I'm not sure what comment you are referring to, but local control and privacy have always been a key aspect on my channel.
@@TheHookUp that's awesome thanks, so many people ignore that..its great you don't..look forward to more videos..
Thanks for this!
The ability to combine this with other capabilities like ultrasonic and LIDAR would be even more next level... Especially as a confidence indicator. MQTT is pretty cool too. Here's a question - ever pick up any used thermal cameras and try to make them work?
The trigger setup page in blue iris looks VERY different now?? It's a bummer there is no update on how to get it working on the latest version. I can see the AI kicking in and checking, but no matter what, no alerts are recorded so SOMEHOW, every time I walk in front of the camera, it refuses to CONFIRM it.....
Dude! you cant be married AND have time to tweak and set all this up! Well, unless its your job as The Hook Up! this was a rather thorough walk through. Makes me confident that I will be able to figure out 2 way audio on Blue iris.
Good work
Hey, man! Awesome content! Thank you for the step by step, including hardware spec. What do you think about using RAID for redundancy in you hardware spec ? Will it work with 24/7 record or might slow it down ?
The big thing I miss with CodeProjectAI vs the old AI Tool approach is the ability to have BI share the AI workload across different servers. I really valued being able to spread a swarm of motion detects across a couple of different servers that would otherwise be idle.
You can share specific cameras to specific servers, but you are correct, there’s no load balancing.
If you click control + double click the alert it goes directly to the ai image info from the alert.
Thank you so very much Rob. Your videos are great guidance. It’s a lot of effort to put all this clear information together, thank you so much.
BTW
Can one camera do license plates and face recognition at the same time?
I have a street view camera that can also see the house entrance.
I have a Nest Hello doorbell but I was not able to connect it to Blue IRIS and so I thought to use the street view camera for face recognition.
Some plate rec cameras from the likes of Axis actually apply an IR-Pass filter in night mode to filter out light from headlights and plate lights, it helps a lot with ensuring the only light present is from the camera’s own IR
Definitely. Purpose built ALPR cameras are always preferable to a DIY solution like this, but not everyone has $3000-5000 to spend on a single camera.
Another great video. I really appreciate your videos. Thank you. Do you have any experience with adding Wyze cameras to Blue Iris using the Wyze-Bridge? Since Wyze no longer offers the RTSP firmware a video like that would help a lot of us.
This video has been a huge help! We added a 1650 per your recommendation on our 24 camera setup. In the CodeProjectAI dashboard I've verified that the GPU is now being use for LPR, object detection and facial detection.
Just one point to clarify: In the global camera settings we set Hardware accelerated decode (restart cameras) to "Intel + VPP". Then on the Camera level for each camera the “Video decode” was left at “Default (Any GPU).”
Are these the proper settings so the cameras takes advantage of our new 1650? You brushed by it kinda quick on the video (loved this instructional format) so I just want to make sure.
Note we have a 10th gen i7.
Great video as always. I think I know the answer, but anyway this will work on Ubiquiti cameras over RTSP?
best video yet... for BI and AI, LPR,etc. A problem I had was after getting schedule using web command, the camera won't respond to PTZ ZoomIn, ZoomOut requests now. Any idea how to fix this? Thanks again for a very informative video.
Great Information and fast moving (which is good). The link for the computer is only an i5, your video recommends an i7. Do you have a link for an i7?
Rob, effectively you make amazing video about camera surveillance.
Just wondering if you ever tried to Hoop up a Coral TPU to your setup. If so, were the results satisfactory ?
Excellent work. I'll bet a video like this take a ton of work to produce.
My only question is: Why did you uncheck 'Default object detection' on the AI tab in Settings?
Another great video, thank you. I noticed in my setup I have Intel+VPP, but I still only get an "I' not an "I+". I have all the setting on Intel+VPP (default and per camera) and the correct Intel Gen CPU too.
I vs I+ depends mostly on what codec is sent from the camera. As long as it says one or the other you’re good.
QUESTION: What monitored home security system do you recommend? LOVE your content and have been watching since your first Shelly video. Thanks!!
I have a self monitored system using Home Assistant.
Great video with a lot of good information. That being said, I would be cautious about setting up CodeProject AI, specifically the custom models for each camera. Read the doc if you don't understand the settings and initially set the custom models value for each camera to "none". If you just start adding custom models without understanding how they work, you can bring your Blue Iris service (and your computer) to its knees, and you will have no clue as to why. Also, my CodeProject AI Object Detection (YOLOv5 6.2) module failed to install correctly, and I was missing several files. If this happens you can fix the problem by uninstalling and reinstalling the module from the Install Modules tab on the CodeProject AI dashboard.
This comment was a sanity saver! I had been running CodeProject in docker in a different VM for a while but decided to run it in Windows like Rob did in the video. Using the default settings and custom models in windows made the BI server completely unusable after that. The Intel video driver would crash repeatedly and Blue Iris would in turn crash. I'm sure if I got intimately familiar with all the specifics of CodeProject I could figure out what was causing it but I decided to just go back to running CodeProject in docker on the other VM as that seemed much more stable for me. Thanks for posting this warning!
@@cmille34 I ran into the same problem immediately because of the default settings. The phone app was balking and not displaying the CPU, ram and disk usage. Once I logged into my Blue Iris server I noticed the CPU usage was bouncing all over the place and maxing out at times. After I set the custom models to "none" for all the cameras, the performance returned to normal. After that I started tweaking the custom models and settings for each camera, one at a time. That's why I would recommend only changing one camera at a time until you're satisfied with the performance. I'm still tinkering with them in order to get the right balance. Although I have the license plate and facial recognition modules installed, I've shut off the functionality in BI until I'm satisfied with the general detection operation.
Do you mean that you had all of the models running simultaneously and that blew up your CPU/GPU usage, or that one of the models specifically killed your performance?
@@TheHookUp After I installed CodeProject AI and resolved the Object Detection (YOLOv5 6.2) module issue, I configured a single camera with a custom models value of "ipcam-combined". The custom models value of all my other cameras was blank. If I am understanding the doc correctly, when it's blank it's using *all* of the models by default (yikes!). You mentioned this in your video as well, but the impact is quite severe if you have a number of cameras. Anyway, this is when I was encountering the CPU/GPU maxing out issue. It wasn't until I began troubleshooting and setting the custom models for my other cameras to "none" that the performance issue began to subside. IMO, defaulting to all models before even configuring a camera is a design flaw that can take down BI and render it difficult to fix. In any event, this isn't a reflection on your video, which is an awesome amount of information in a short period of time, but just something I wanted to mention to others who may get tripped up by this design issue.
Gotcha. Yes, the reason I did things in the order that I did was so that you would work with a single camera and then clone those settings into the rest of your cameras as they get added.
Hey Rob! Thanks for update the content about BI.
Can i ask you something? You should make a in deep video noob friendly about BI Mqtt + HA.
Thanks again, man.. greetings from Brazil 🇧🇷
Wish i could hire you for a day haha
Ok, got everything setup. Love the quick format of your video (maybe a bit too quick, thanks RUclips for pause!) and you touching on all the necessary/important steps from soup to nuts. Very helpful in getting other things configured besides Codeproject. Have 4 of my 12 cameras feeding into CP and it's working great.
One minor issue is the resolution of the "new face" images. Are they supposed to be so tiny? I haven't tried to create known faces with them yet.
"new faces" should just be cropped faces from your camera's normal resolution. So if the face was a small portion of the image it will be a small overall file. If it seems REALLY small it's possible it could be pulling from your substream, not sure what to do about that.
20:37 hide cancelled alerts isn't available on current version anymore and replaced with "new alert/analysis on re-trigger after cancellation", should I uncheck it as well? The vid is a lot of information in 20 minutes I've spent 6hrs to reach here.
Rob I really love the idea of this setup but I have minimal computer knowledge. Is this something someone with basic computer skills can handle by following your written guide or is this best left up to a professional? Also, you’ve recommended the Annkec800 in the past with BI, are the cameras you listed in description the ones you now prefer? Haven’t seen an update on PoE cameras… thanks!!
Thanks for your great content. I have a question regarding Reolink RLC-810A and Blue Iris. We have a situation where I can't do any port forwarding on the site where two Reolink RLC-810A are meant to be installed (Reolinks are behind a cisco meraki mx85 firewall and port forwarding is not allowed / not possible at all). uPNP haven't been tried against the MX85 yet, I will. However: On the other side I have the Blue Iris-installation machine which will take care of everything with the Reos, but that too is also behind a ISP using a "public ip for many clients" so I don't have a public IP right now but actually applied for a public one tonight. The Blue Iris machine has a RLC-811 today on it's internal network, no problems, but I want to add these RLC-810A:s soon.
My question is: Is it possible to connect Reolink cameras to Blue Iris (with both sub- and mainstream, will be using codeproject.ai) when both sides are behind firewalls? I know Reolink has the UID to connect to their app - but that doesn't make it to Blue Iris the way I see it. Any recommendations? If it's way too difficullt then I will put a machine specifically at the Reos location for VPN-tunneling the Reo traffic to a dedicated server (with Softether VPN Server) that I have hosted, and then make the Blue Iris machine connect to that same Softether VPN-server aswell. Thank you for reading and keep up the excellent work! Sincerely, Sea.
I'm just about to setup my security system at my new home and after rewatching your other videos I realized that you no longer recommend the Reolink rlc810a with Blue Iris. Can you give me your new best recommendation for best value hd camera?
Hi Rob, yet again another great video with detailed instructions. I noted before my DeepStacks stopped pushing notifications and as soon as I switched over to CodeProject everything worked again. Only question I have now is under the pre-trigger record time within "alerts" you set yours to 2secs (around the 13:08 mark). I have all my setting the same per this video, but when I went back it is set to '5sec'. I changed back to 2, logged back in and still gets switched back. Any thoughts on why this may be happening? Thank you again for the great video. I will go back and set up plate detection after I let this setup run to tweak.
Hey Rob, tips for those of us that have hit the faces dialog? I'm not getting any new faces saved. Great video by the way!
Great video, Rob! I think the only miss here is overpaying for hardware and power instead of using a Coral TPU with codeproject, or Jetson nano with Deepstack. $100 takes the place of the highest end video card.
I don't think Coral is supported with CP.ai yet, is it?
Yes! It is now supported!
@@scottswf238 I looked last night and it seems like it's still in very early stages with inference times in the 100-300ms, and has to be run on a raspberry pi?
it runs in docker on a raspberry pi. if you choose smaller models it runs faster, so with the small model i reliably get 40-80ms detection times. there’s a smaller model called tiny that should be faster than that.
I’m not sure if it supports custom models yet though, and i don’t think it supports facial recognition/lpr, so you’d still want to use gpu for that
@@michaelwozniak2850 That does not sound bad! Especially for the cost and power. I use s Jetson nano and Deepstack. It’s doing dozens of images per minute, due to a high traffic zone. I’d love to move to Codeproject and I hope they support this platform soon also. I can’t justify buying a full-fledged video card versus this set up.
@TheHookUp
Hi, I have a really dumb question. Would it be worth the trouble of trying to adapt this system into a vehicle? The Dahua camera looks like a (an expensive) professional solution. Thank you!!
Thanks for this great video. I now use a Synology, but it's a bit limiting. It only uses 63 watts of power. With the current high energy prices, do you have any idea about the average power usage in this setup?
I would assume the hard drives are a significant part of that. At least if you have 6+ harddrives.
If recording continous wouldnt it be better to save the new alerts/triggers in the new folder but have them actually get moved to the Stored folder after some time? Or does continuous record to that stored folder regardless as is? I'd assume the events are saved there in "New" so they are different from continuous, so they should be moved to storage?
Great video. Saw your past videos. I understand that Realink has *now* fixed dual stream setup in Blue Iris. As per the latest product line, which cameras do you use or suggest ?
Also - BI HA Integration v/s Reolink NVR HA integration - which supports more features, like AI Identified a face, or AI found unidentified person ?
great video rob.... i used this when it came out to install... sadly blue iris loves updating all the menus every version and its always a pain to do a year later...
Have been looking all over the web and RUclips on how to get Blue Iris push notifications to work using a VPN. I am using Wireguard through my UDM Pro. Apparently most people use Port Forward, which I know is not recommended. What did you do for your setup? Awesome videos! Thanks!
Hi, great video! Do you have a low energy consumption recommended hardware set up for Blue Iris + Code Project AI? Is it possible to add the quality of graphics card needed to a NUC and if so will this generally be lower power than a standard size PC or will the AI requirements nullify the gains? Any steer would be appreciated - I am about to build a system from scratch.
At 7,07 into the video, the box next to the "used custom models" need to populated, but mine is showing empty, even after a restart. Because of this I can't get it to work. Does anyone know what may have gone wrong?
Great job always, LPR camera are very professional cam to dial the best pics. thanks for recommending our IPC-HFW5241E-Z12E
Amazing video Rob. Is there any way to use the facial recognition triggers to run automations in home assistant? I.e unlock the door when a face is recognized etc
Of course. MQTT can pass face data through to HA just like plate data.
Thanks for the guide. Awesome vids as always. Just confirming that this setup of blue iris and cameras are IP cameras, correct? So it won't really work if someone has an NVR kit setup? Sorry for the rookie question.
Correct, IP cameras.
Great vid Rob! In setting I clicked 'Faces' by mistake. My face recognition is not working. Is there a way to fix the error?
This is freaking awesome! Which Reolink cameras will work with this setup?
5mp cameras with RTMP streams are the only reolink cameras I’d recommend using with blue Iris.
@5:33 - Are there any downsides or upsides to using the other video decoders? Particularly NVDEC or DirectX? For instance I have an older 16-core Xeon workstation that I don't think has Intel VPP, but it's super easy to upgrade the graphics card compared to MoBo+CPU.
Edit: Some discussion already in these comments about NVDEC being experimental. Can anyone confirm that NVDEC is not reliable? DirectX doesn't work either or am I mistaken that DirectX uses GPU?
Great video as always rob. just one question I have been looking for the Json string for the mqtt topic can you post please. keep the great work coming.
{"plate":"&PLATE", "AlertImgPath":"&ALERT_PATH", "Alert_AI":"&MEMO", "Date":"%Y-%m-%d %H:%M:%S","Camera":"&CAM"}
@@TheHookUp as allways thanks
I think it is best to stop blue iris when installing modules in CPAI unless you have a really good PC or very little cameras.
I had issues with modules installing correctly until I stopped BI service.