5.4.1 - April 3, 2021 - Support for DeepStack facial recognition

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,432
Reaction score
47,556
Location
USA
LOL yes I thought you meant installing another camera, yes cloning would work. I'll try that. Thanks for the clarification. Actually that should work really well to separate the road traffic from the traffic entering my driveway regardless of deepstack. When cloning you can use the same stream from the camera and not increase the total bandwidth right?
Yes, cloning does not add any significant CPU load or bandwidth. Go into the main camera and check the clone master. Then copy that camera and make sure that the clone camera does not have clone master checked. The clone camera should then have an * next to it letting you know it is cloned.

Yeah that camera should work fine.
 

tech101

Known around here
Joined
Mar 30, 2015
Messages
1,472
Reaction score
2,125
Location
SF BayArea, USA
Just installed the 5.4.2.0 - Will wait for few alerts to see how the Playback is working if there is any issue like earlier with pre buffer size.
 

Bosty

Getting the hang of it
Joined
Oct 3, 2016
Messages
43
Reaction score
38
Location
Vermont
If I am monitoring my driveway for vehicles how can I do this when there are already vehicles parked in my driveway? Everytime I get an alert deepstack puts it through because AFAIK its detecting the vehicles that are already in the yard.
I have a car parked across the street. I zoned the whole image and when a car went by, it recorded 2 cars (like it should). I then unmasked the car and now when a car goes by, it only records the one car for motion and doesn't acknowledge the parked car. I'm on version 5.4.1.1

Unmasked Parked Car.JPG
 

tech101

Known around here
Joined
Mar 30, 2015
Messages
1,472
Reaction score
2,125
Location
SF BayArea, USA
Update - Seems like Buffer issue is fixed.

Also As Per Update Notes from the Help file seems some really cool improvements.

5.4.2 - April 6, 2021
By popular demand, you may now delay recording or snapshots until alerted rather than
starting immediately upon a trigger. This allows the AI time to analyze the trigger before
recording begins.
With a new option on Settings/AI you may now select to save unknown faces to a specific
folder for later review or import.
At startup and after PTZ movement, DeepStack will now analyze the scene for initial objects
in order to prevent alerts on static objects.
The marked-up alert image is now re-saved to the database so that it will appear on the
alerts list.
With a right-click option in the Viewer, you may run a snapshot through DeepStack for
analysis. This will eventually be expanded to include BVR content.


1617722262666.png

This is cool, This way I now can tell what deepstack is looking at. Now it be nice to have some training at some point built in where if something is not detected we can do multiple passed and make deepstack get better and better :D


1617722394642.png
 

D0T-C0M

Getting the hang of it
Joined
Feb 21, 2021
Messages
103
Reaction score
28
Location
NB, Canada
@Bosty Nice to know that masking out a vehicle works in deepstack but using just one camera I cannot mask my vehicles because the vehicles in my driveway are never parked in the same spot and when they are not there I don't want to mask that area. Using a cloned camera will work in my case as I can use different masking zones, just not sure deepstack will be reliable enough yet. That is what I'm testing.
 

fenderman

Staff member
Joined
Mar 9, 2014
Messages
36,897
Reaction score
21,250
@Bosty Nice to know that masking out a vehicle works in deepstack but using just one camera I cannot mask my vehicles because the vehicles in my driveway are never parked in the same spot and when they are not there I don't want to mask that area. Using a cloned camera will work in my case as I can use different masking zones, just not sure deepstack will be reliable enough yet. That is what I'm testing.
You dont need to mask the vehicle. BI detects that the vehicle has not moved since the last analysis.
 

Bosty

Getting the hang of it
Joined
Oct 3, 2016
Messages
43
Reaction score
38
Location
Vermont
So there is no need to have zones anymore and just turn it off under motion?
 

Bosty

Getting the hang of it
Joined
Oct 3, 2016
Messages
43
Reaction score
38
Location
Vermont
You dont need to mask the vehicle. BI detects that the vehicle has not moved since the last analysis.
So I turned off the mask and every alert now includes the parked car along with whatever object is in motion. So it is not ignoring static objects...

By popular demand, you may now delay recording or snapshots until alerted rather than
starting immediately upon a trigger. This allows the AI time to analyze the trigger before
recording begins.
What setting is this to delay the recording?
 
Joined
Jan 3, 2019
Messages
17
Reaction score
9
Location
Germany
So it is not ignoring static objects...
I have the feeling we are in an early stage... on 5.4.0 I had the occupied-symbol on my test-scenario, on 5.4.1.. it "redected" always where it has been occupied before in the same situation. Maybe write him a bugreport or wait some days, I think that behaviour will be refined. It is a heck of new stuff which is implemented the last days. There will be some quirks at the moment.
 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,690
Location
New Jersey
OK. I'll plead advanced age, being "set in my ways" and some dumbness. What is the big trick to getting the UI control panel for DeepStack to run in Win10? I downloaded the UI .zip and unzipped it. I read the entries on their forum about it, and everything seems geared to using it in a Linux docker. I don't want to learn Linux or run docker containers. So I guess it's actually two questions - 1-Dooes the UI require Linux and therefor a docker in Win10 and 2-What's the magic to make the UI work in Win10 if it doesn't require Linux in a docker?
 

D0T-C0M

Getting the hang of it
Joined
Feb 21, 2021
Messages
103
Reaction score
28
Location
NB, Canada
No you download deepstack for cpu or gpu from the link in the BI help file. You just install it and BI will configure it in the global settings and then you enable it in each camera that you want it on.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,432
Reaction score
47,556
Location
USA
OK. I'll plead advanced age, being "set in my ways" and some dumbness. What is the big trick to getting the UI control panel for DeepStack to run in Win10? I downloaded the UI .zip and unzipped it. I read the entries on their forum about it, and everything seems geared to using it in a Linux docker. I don't want to learn Linux or run docker containers. So I guess it's actually two questions - 1-Dooes the UI require Linux and therefor a docker in Win10 and 2-What's the magic to make the UI work in Win10 if it doesn't require Linux in a docker?
I just downloaded the CPU version here and installed it and then opened BI and it worked. I did not open or try to run Deepstack, just let BI do it. I am not running dockers or Linux.

 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,690
Location
New Jersey
I have DeepStack running with no problem and have used PowerShell to start it as well as letting BI start it. There is a GUI for DeepStack, itself, that I'm trying to get running.
 

shannondalebreaux

Getting the hang of it
Joined
Jun 6, 2018
Messages
82
Reaction score
29
Location
louisana
Update - Seems like Buffer issue is fixed.

Also As Per Update Notes from the Help file seems some really cool improvements.

5.4.2 - April 6, 2021
By popular demand, you may now delay recording or snapshots until alerted rather than
starting immediately upon a trigger. This allows the AI time to analyze the trigger before
recording begins.
With a new option on Settings/AI you may now select to save unknown faces to a specific
folder for later review or import.
At startup and after PTZ movement, DeepStack will now analyze the scene for initial objects
in order to prevent alerts on static objects.
The marked-up alert image is now re-saved to the database so that it will appear on the
alerts list.
With a right-click option in the Viewer, you may run a snapshot through DeepStack for
analysis. This will eventually be expanded to include BVR content.


View attachment 86194

This is cool, This way I now can tell what deepstack is looking at. Now it be nice to have some training at some point built in where if something is not detected we can do multiple passed and make deepstack get better and better :D


View attachment 86195
What kind of camera are you using here that 8MP the image looks super clear? Is it a PTZ as well I see the writing in the left corner.
 

tripp396

Getting the hang of it
Joined
Jun 18, 2020
Messages
65
Reaction score
30
Location
Minnesota
So I turned off the mask and every alert now includes the parked car along with whatever object is in motion. So it is not ignoring static objects...

By popular demand, you may now delay recording or snapshots until alerted rather than
starting immediately upon a trigger. This allows the AI time to analyze the trigger before
recording begins.
What setting is this to delay the recording?
I am seeing the same behavior on 5.4.2. Parked cars are being included, not ignoring static objects.
 

anijet

Pulling my weight
Joined
Mar 10, 2014
Messages
346
Reaction score
165
Location
CA
............. When cloning you can use the same stream from the camera and not increase the total bandwidth right?
That is correct.

Just copy/export the existing camera, add a new camera, import the copied one and change the camera name and short name before you save it. Then modify the settings from there.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,432
Reaction score
47,556
Location
USA
so DS really knows the difference between car, boat, deer, raccoon, possum, human, skinny human?
here is the evergrowing list LOL:

person, bicycle, car, motorcycle, airplane,

bus, train, truck, boat, traffic light, fire hydrant, stop_sign,

parking meter, bench, bird, cat, dog, horse, sheep, cow, elephant,

bear, zebra, giraffe, backpack, umbrella, handbag, tie, suitcase,

frisbee, skis, snowboard, sports ball, kite, baseball bat, baseball glove,

skateboard, surfboard, tennis racket, bottle, wine glass, cup, fork,

knife, spoon, bowl, banana, apple, sandwich, orange, broccoli, carrot,

hot dog, pizza, donot, cake, chair, couch, potted plant, bed, dining table,

toilet, tv, laptop, mouse, remote, keyboard, cell phone, microwave,

oven, toaster, sink, refrigerator, book, clock, vase, scissors, teddy bear,

hair dryer, toothbrush
 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,690
Location
New Jersey
I'm still seeing some time discrepancies with 5.4.2.0. I was just scanning clips in UI3. The clip animation starts at the time of the actual clip being moused over then jumps back to the previous clip. Not a real big deal, but still a little weird.

Nobody has an answer on the DeepStack UI question?
 

tripp396

Getting the hang of it
Joined
Jun 18, 2020
Messages
65
Reaction score
30
Location
Minnesota
I'm still seeing some time discrepancies with 5.4.2.0. I was just scanning clips in UI3. The clip animation starts at the time of the actual clip being moused over then jumps back to the previous clip. Not a real big deal, but still a little weird.

Nobody has an answer on the DeepStack UI question?
I may be wrong but last I saw the UI version for windows needed an activation key, and deepstacks login page wasn't work (last I checked).

The link here: has a windows exe that will install deepstack, then it's as simple as opening a command line and passing something like:

Code:
deepstack --VISION-DETECTION=True --PORT=80
EDIT: I saw your other message, realized you have the commandline version running but want GUI. Sorry about missing that part of it.
 
Top