Look for 2 pieces of SW - 1.) Auto tagging and 2.) Auto isolate frames to delete

[A_S]

Hi

I've found that moving to a quicker drive remote and electronic shutter on my main camera - I am ending up with 1000s of images after an event and I am looking for some sort of work flow solution to

A.) Isolate images of a specific car quickly (tagging)
B.) Delete any OOF images quickly without going through them - as my remote obviously records a hell of a lot of frames that need to be deleted, plus if I am doing low shutter speed stuff down at 1/20s naturally I have a lot of crap to clear before and after the main frame I am shooting for. (auto isolate images to delete)


Regarding A:

I can't seem to find (google) any software that will block analyse a set of images and tag each image with a Name or Number (each car has a number which is often visible), I am also thinking there must be a way for AI to isolate all images of a certain car based on livery/shape/model and tag it with a specific code so i can pick out each car individually, or at least each model.

Taking that a step further - surely if it can pick out a car number from a side perspective, then sees the same car latter in the image sequence but at a 3/4 angle (but can't see the number) AI should be able to recognise it as the same car and tag it with that number even if the number is not visible.

I know this won't be 100% accurate but will surely save me the hours I spend at the moment trying to isolate images of specific cars.

Any thoughts?

 

[Box Brownie]

None too sure but perhaps look at Excire Foto.

Excire: AI Image Management Software for Mac & PC

Excire - viele intelligente Funktionen unterstützen Sie dabei, Ordnung zu halten und die Fotos zu finden, die Sie gerade suchen.

excire.com

They in common with all software have a trial period.

 

[A_S]

The 2022 version that acts as a lightroom plugin might be a good shout - Thanks.

Not sure about the 2024 version as it looks like a stand alone program which requires you to use it to find images - I'll do a bit of research this weekend

 

[A_S]

I'm surprised no one else has any other solutions? Surely there must be a lightroom plugin out there somewhere...

my quick searches brought something up called wordroom - but that requires individually tagging every photo so not practical for 1000s of images

although this might be more promissing

Any Vision – John R. Ellis Lightroom Plugins

Automatically tag Lightroom photos with objects, activities, landmarks, logos, dominant colors, face expessions, and embedded text (OCR). Uses Google AI.

johnrellis.com

 

[A_S]

None too sure but perhaps look at Excire Foto.

Excire: AI Image Management Software for Mac & PC

Excire - viele intelligente Funktionen unterstützen Sie dabei, Ordnung zu halten und die Fotos zu finden, die Sie gerade suchen.

excire.com

They in common with all software have a trial period.

Tried this today, it neither picked up the registration or number. seemed to just apply a generic set of keywords so not sure of the benefit to me at least.

 

[Box Brownie]

Tried this today, it neither picked up the registration or number. seemed to just apply a generic set of keywords so not sure of the benefit to me at least.

A pity.......but worth a try?

 

[A_S]

A pity.......but worth a try?

absolutely - The other one I found above - works really really well. it picks up any text or number such as registration, car number or sponsor and adds it to the meta data.

I just need to work out whether google pricing (as it uses google's ai engine) is efficient enough to use it on 5000ish photos an event for 20 events a year....

Also tried photo mechanic for culling as I couldn't find a suitable lightroom plugin. Also not really quite right for cars. of my 3 1000 image test samples, the most it picked up was 30 -70 "green or yellow" flagged images, the rest it left un flagged so not a good solution for the culling element.

 

[JohnMcL7]

Unfortunately I don't have any suggestions but surely what you've specified must exist, I've taken part in cycling events where there's huge amounts of people taking part yet within 24-48 hours of the race finishing photographers have images uploaded and tagged with the race number. I've been having a look but not found anything suitable though.

 

[A_S]

Unfortunately I don't have any suggestions but surely what you've specified must exist, I've taken part in cycling events where there's huge amounts of people taking part yet within 24-48 hours of the race finishing photographers have images uploaded and tagged with the race number. I've been having a look but not found anything suitable though.

john, this is the best I've found but relys on google AI - for which there is a cost ($1.50) per 1000 photos after the first 1000 each month.... thats only for the OCR bit (text recognition). It cal also tag landmarks (e.g big ben or eiffel tower), facial expressions, GPS co-ords too I beleive - but each "module" has the same pricing structure, so if you applied all 7 modules each time over 1000s of photos costs would mount.

Any Vision – John R. Ellis Lightroom Plugins

Automatically tag Lightroom photos with objects, activities, landmarks, logos, dominant colors, face expessions, and embedded text (OCR). Uses Google AI.

johnrellis.com

I've applied it to some images and it works well for my purposes of isolating specific cars in folders of 1-4000 images.

I just wish there was one with a one off charge rather than relying on the google pay bit.

Its useful but makes the second bit more important to reduce the costs - i.e the culling bit before you run the photos through the google AI processing

 

[Pound Coin]

A sports set up I knew sent the image files to Asia to be tagged. I guess this will be something for AI eventually.

 

[garryknight]

I guess this will be something for AI eventually.

It's already here. I got the Bing Chat AI to look at my photo here and told it to "suggest suitable hashtags for this image". It came back with this:

Some possible hashtags for this image are:

- #carnegie #mural #art #history
- #blackandwhite #photography #vintage #culture
- #booklover #reading #inspiration #education

You can choose the ones that best suit your purpose and style. I hope you like them.

I'm not so sure about #carnegie (it was outside the National Gallery in London), nor #booklover or #reading, but the rest are OK. The main problems with this method are: 1) having to do them one at a time, manually; 2) the two minutes it took to analyse just one image; and 3) that it didn't suggest the obvious (to me, at least), #woman, and #hands.

 

[Pound Coin]

It's already here. I got the Bing Chat AI to look at my photo here and told it to "suggest suitable hashtags for this image". It came back with this:



I'm not so sure about #carnegie (it was outside the National Gallery in London), nor #booklover or #reading, but the rest are OK. The main problems with this method are: 1) having to do them one at a time, manually; 2) the two minutes it took to analyse just one image; and 3) that it didn't suggest the obvious (to me, at least), #woman, and #hands.

Yes, but that isn't tagging a series of photos with the registration number, which is what I was suggesting.

Looking at your AI generated list of tags it's not far off what a Xitterati might have put on one of their posts, whilst nodding sagely.

 

[Craikeybaby]

Unfortunately I don't have any suggestions but surely what you've specified must exist, I've taken part in cycling events where there's huge amounts of people taking part yet within 24-48 hours of the race finishing photographers have images uploaded and tagged with the race number. I've been having a look but not found anything suitable though.

What is taking them so long? When I shoot races I make sure I have the pictures online that evening. The AI tagging does look helpful, although I don't think the site that I upload to would use the keyword, so I would still need to do the tagging after, which isn't really that much hassle.

 

[JohnMcL7]

What is taking them so long? When I shoot races I make sure I have the pictures online that evening. The AI tagging does look helpful, although I don't think the site that I upload to would use the keyword, so I would still need to do the tagging after, which isn't really that much hassle.

I assume because of the number of cameras and staff then returning to another country, that takes some time to get the photos together.

 

[Pound Coin]

What is taking them so long? When I shoot races I make sure I have the pictures online that evening. The AI tagging does look helpful, although I don't think the site that I upload to would use the keyword, so I would still need to do the tagging after, which isn't really that much hassle.

How many participants are you photographing?

 

[Craikeybaby]

It is in the hundreds, rather than thousands. I am by no means the quickest though, there are usually images already uploaded before I've got home and downloaded all my images.

 

[A_S]

Just to clarify there ways to isolate images to use in the field for example rating, then applying a standard preset to that and exporting, uploading, job done.

But when X driver or Y co-drive asks you to send them proofs of car X - I've got

A.) A main camera and remote, so 2 cameras
B.) Each which take between 1000-4000 images (the remote more of course as I don't have total control/visibility over it so it bursts more)
C.) 2 runs per day on a short rally, a 2 day event or with more runs/stage visits

So in essence you are looking for a specific car in a pool of between 4000 and 8000 images.

Of course there are ways to shortcut such as looking at running order and previously processed photos of cars in and around that specific car, but cars go out of order, running order changes, re-groups happen and one car which comes through as say the 50th first time might come through as the 60th next time.

It makes finding a specific car very time consuming... Hence wanting a solution that solves the endless task of looking for images. I am not quite sure your scenario is too similar Craikeybaby - or at least it doesn't seem so anyway.