Software Engineering Courses (SECourses)•3y ago

I have prepared a folder with handpicked classification images, but when I start training dreambooth

I have prepared a folder with handpicked classification images, but when I start training dreambooth ignores them and starts creating them. Have you had this problem? Thank you in advance for any help.

YYVD I have prepared a folder with handpicked classification images, but when I start...

Furkan Gözükara SECourses•4/27/23, 10:39 AM

must be a configuration error

Furkan Gözükara SECourses•4/27/23, 10:39 AM

after saving config click load

Furkan Gözükara SECourses•4/27/23, 10:39 AM

you tried that?

Furkan Gözükara SECourses•4/27/23, 10:39 AM

also show me settings please

YVDOP•4/27/23, 10:42 AM

Thank you, let me check

YVDOP•4/27/23, 10:47 AM

YVDOP•4/27/23, 10:47 AM

Ive tried saving and loading with same issue

Furkan Gözükara SECourses•4/27/23, 10:47 AM

show settings please

YVDOP•4/27/23, 10:47 AM

What is interesting is that it recognizes the classification images it generated on its own, but it wont recognize the ones I put in

Furkan Gözükara SECourses•4/27/23, 10:48 AM

ok use

Furkan Gözükara SECourses•4/27/23, 10:48 AM

pre processing tool

Furkan Gözükara SECourses•4/27/23, 10:48 AM

in train tab

Furkan Gözükara SECourses•4/27/23, 10:48 AM

it will fix

YVDOP•4/27/23, 10:52 AM

Great thanks! Do I have to use a specific ratio for this?

YVDOP•4/27/23, 10:53 AM

I intend to use a 960 x 512 when working. On the preprocess images it says 512x512

YYVD I intend to use a 960 x 512 when working. On the preprocess images it says 512x5...

Furkan Gözükara SECourses•4/27/23, 10:56 AM

your classification images probably should be same dimension

Furkan Gözükara SECourses•4/27/23, 10:56 AM

as your training images

Furkan Gözükara SECourses•4/27/23, 10:57 AM

if you dont change max resolution they will get downscaled

YVDOP•4/27/23, 10:57 AM

great, thanks a lot!

YVDOP•4/27/23, 11:00 AM

keep up the great work

YVDOP•4/27/23, 11:06 AM

It worked! im so happy!

rafraf•4/27/23, 11:10 AM

Hey @Furkan Gözükara SECourses are you aware of any GPT assistants who can already help with captioning datasets. I find myself constantly going back to make changes to a tag word or a string, and have to go back to all the dataset txt files, and it's quite time inefficient. Would imagine being able to leverage GPT + python would bring captioning from an hours to minutes. Any luck yet with applying these GPT agents to captioning?

Rrafraf Hey @Furkan Gözükara SECourses are you aware of any GPT assistants who can alrea...

Furkan Gözükara SECourses•4/27/23, 11:22 AM

there are already blip and other captions

Furkan Gözükara SECourses•4/27/23, 11:22 AM

with the caption thing what is important is not describing that image

Furkan Gözükara SECourses•4/27/23, 11:22 AM

the important thing is which words you want to improve

Furkan Gözükara SECourses•4/27/23, 11:22 AM

you want to do fine tuning right?

FFurkan Gözükara SECourses there are already blip and other captions

DeadMan1999•4/27/23, 11:28 AM

although BLIP makes the process much faster, in my opinion it is less accurate. Sometime it produces wrong descriptions for the images. This might affect the training right? I have not tried GPT-4's image description feature yet, but from what I heard, it is quite accurate. maybe a comparison between BLIP and GPT-4 is needed

FFurkan Gözükara SECourses there are already blip and other captions

rafraf•4/27/23, 11:31 AM

Yes precisely. I am doing Lora of recognisable fashion model person wearing recognisable fashion outfit. I need to be able to trigger the recognisable face and clothes, so it is rigid, and to do that I am using key tags in the caption to make the ai aware what is in the picture. I havent optimised yet for face, i am working on outfit to be faithfull to my studio original photo reference that I shot. But the point is, that BLIP captioning only creates the txt structure. The caption guess is pretty useless. But anyway, what i want is an Ai assistant to propagate the editing. So i decide, i need to change a term, say about the buttons, i need the button tags to change across all the txt files in the dataset. Do you know what I mean.

rafraf•4/27/23, 11:31 AM

I am just asking about an Ai assistant to go in to each txt file and change them. I am using 50 images per dataset concept, and i have about 10 concepts (outfits). So that's a lot of txt files to recursively edit.

rafraf•4/27/23, 11:32 AM

So I would imagine a python code would be involved to delegate text editing across all text files...

FFurkan Gözükara SECourses you want to do fine tuning right?

rafraf•4/27/23, 11:32 AM

I want to fine tune yes.

DeadMan1999•4/27/23, 11:32 AM

so you want to replace a tag with another tag and do the same for all the text files?

DDeadMan1999 so you want to replace a tag with another tag and do the same for all the text f...

rafraf•4/27/23, 11:33 AM

Yes. Say, "white button" to "4 white patterned button". There are 50 images with those buttons. Would be great for the Ai assistant, to propagate the change across all the txt files

rafraf•4/27/23, 11:34 AM

CLIP captioning only guesses what things are. I need something to help me edit details

rafraf•4/27/23, 11:34 AM

So yes, replace tags

rafraf•4/27/23, 11:34 AM

--across a group of txt files

DeadMan1999•4/27/23, 11:34 AM

you could download notepad++ and use "find and replace in files" feature, it will find and replace words in all the files you have mentioned

rafraf•4/27/23, 11:35 AM

oh i did not know that. It can do this already

rafraf•4/27/23, 11:35 AM

How silly of me

rafraf•4/27/23, 11:35 AM

Ok, will give it a try now

DeadMan1999•4/27/23, 11:36 AM

yes, I learned this 2-3 days ago as well, maybe from @Furkan Gözükara SECourses videos or from other dreambooth tutorials on YT

rafraf•4/27/23, 11:38 AM

As an extension though, It would be ideal if an Ai assistant could be fine tuned to execute things in my captioning categories. I always say, "head turned to side" or "head tilted". etc. Wouldnt an Ai assistant be ideal to learn what you like to caption and do it for you? But i suppose this is still some ways off

DeadMan1999•4/27/23, 11:40 AM

ctrl+f > find in files > "input directory where files are present", in Filters put: ".txt"

DeadMan1999•4/27/23, 11:40 AM

Rrafraf As an extension though, It would be ideal if an Ai assistant could be fine tuned...

DeadMan1999•4/27/23, 11:41 AM

yes, this could be possible

DDeadMan1999 yes, this could be possible

rafraf•4/27/23, 11:42 AM

maybe with MiniGPT?

DeadMan1999•4/27/23, 11:44 AM

ohh, miniGPT do have Image describing feature

FFurkan Gözükara SECourses yes you can do that

Khang•4/27/23, 11:49 AM

I'm using Runpod

Khang•4/27/23, 11:49 AM

Does anyone have a problem where Dreambooth Load Setting doesn't load and keeps loading forever? I tried to load a 100-epoch checkpoint to train it until 150 epochs, but it keeps loading. I have faced this problem several times in the past, but I wasn't sure how to fix it. I have restarted SD multiple times sometimes the error goes away on its own, sometimes it doesn't.

Rrafraf I want to fine tune yes.

Furkan Gözükara SECourses•4/27/23, 11:50 AM

ok you want to fine tune which words?

I have prepared a folder with handpicked classification images, but when I start training dreambooth

Similar Threads