A DeepStack custom model for detecting 352 common logos - GitHub - OlafenwaMoses/DeepStack_OpenLogo: A DeepStack custom model for detecting 352 common logos
Quoting myself to add the following responses from Ken...I sent an email to Ken with these proposals to help when we’re debugging DeepStack performance... (no response yet)
1. Would it be possible to make the 'Testing & Tuning > Analyze with DeepStack' option reflect the entries in the 'Custom models' box for the active profile?
2. And/or could the 'Testing & Tuning > Analyze with DeepStack' option display a popup with the same info as the &JSON macro for alerts?
The Testing & Tuning > Analyze with Plate Recognizer option currently displays the json response in a pop-up. A bit rudimentary but it works. Hopefully, your request for Deepstack json output isn't too far away.Quoting myself to add the following responses from Ken...
1. Using camera's current settings for objects with 188.8.131.52. (6/11)
Note: the emphasis is mine; I've asked if the camera's current settings will also work for faces & custom models.
2. I like the idea generally about displaying the raw JSON simultaneously. I would need to work on that as a feature suggestion thanks. (6/8)
Lucky LOL.OK. I just checked and got a logo capture, at least it captures the logo using "tuning". It did identify the FedX logo very early in the clip and also identified that it was a truck. The clip is actually listed as a truck and not FedX which is a little disconcerting but I guess that's because the confidence level is higher for the truck. I didn't notice any blue border, either.
View attachment 92089
That has been my experience as well. Not sure where there is a chair on my stairs, but yeah, it seems to tag anything it finds in analyzing. No complaints though.I believe playback will show all of the items DeepStack can identify? At least it always has with mine LOL. I am getting potted plants, toothbrush, LOL.
The images for several of my cams seem to be target-rich environments, because I keep getting 6-10 overlapping rectangles and labels. Often this yields a less-than-helpful marked-up image because the tag I’m trying to verify per my settings is obscurred a “potted plan” or a “boat” or … I like seeing what’s going on under-the-hood, but would prefer “seeing” it with the &JSON output.I believe playback will show all of the items DeepStack can identify? At least it always has with mine LOL. I am getting potted plants, toothbrush, LOL.
I use it for most but a couple it's just not practical.. this camera here for instance is probably best flawless with alerts and sending MMS trigger pictures. Memo might not always be accurate but I never miss a person or vehicle coming in and it's quicker to see a text than open bi .. I'm extremely happy with this particular setupI have sit back for the most part and watched this evolve. I finally took the leap yesterday and installed Deepstack and set it up, on 4 cameras.
Results have been very mixed, I can only get it to work on two of the four cameras reliably. And by work, I mean deepstack actually working to record a clip.
The two cameras I have issue's with, both show deepstack doing it's job when using test and fine tune, but for the most part, deepstack doesn't do anything when just letting it do it's thing.