Hi everyone,
I’m a photographer and I usually capture hundreds of images during events. I’m looking for an n8n workflow or template that can automatically analyze and sort my photos into categories such as:
-
Images with closed eyes
-
Blurred or out-of-focus photos
-
Duplicate or near-duplicate images
-
Photos with lighting issues (too dark, overexposed, etc.)
Is there an existing template or recommended setup to achieve this?
Any suggestions, examples, or workflows would be really helpful. Thanks!
And i m not the n8n expert 
so readymade template really appreciated.
Thanks in advance
Mukul O.P. Kalra
1 Like
N8n should be used to create automated workflows that integrate with Amazon Rekognition, a cloud-based service from AWS that uses deep learning to analyze images and videos.
and what I think Google Nano banana can also do the same
Hey @MukulOPKalra !
Here you can find a free workflow, so you can play and adapt to your needs!
Give it a try, and if you get stucked post here errors you encounter.
Have fun!
Dear All,
I found this use case really interesting, so I went ahead and tried to build the workflow myself. However, I ran into a few issues and would greatly appreciate it if any experts here could point me in the right direction.
-
How to properly pass a photo to an LLM for analysis? Currently, I’m using a rather clumsy workaround: I download the image from Google Drive, upload it to ImgBB to get a public URL, and then feed that URL to the model. Is there a cleaner/more reliable way to do this?
-
Issues with image analysis in Gemini models Gemini 1.5 Flash doesn’t seem to analyze the photos correctly. I also tried the “Gemini 1.5 Flash Image” variant , but the results are still inaccurate.
-
What is the current best practice for building a robust image-analysis workflow like this?
Any tips, recommended tools, or alternative models that handle images more reliably would be very helpful!
Thank you in advance!
Workflow Plan:
Workflow Run:
Workflow:
1 Like
After studying other people’s templates, I revised the workflow with the following improvements:
-
Instead of uploading photos to ImgBB, the photos are now processed locally, converted to a Base64 string, and sent directly to the LLM. This is much cleaner and more efficient than the previous method.
-
I tested several models and found that MiniMax M2 gives the closest results to what I expected. The only mistake was with a closed-eyes photo (which should have gone to the Discard folder) being incorrectly sent to the Review folder. Otherwise, underexposed photos were correctly routed to Review, and good photos were accurately placed in the Good folder.
Feel free to share any suggestions or feedback if you have ideas for further improvements!
1 Like
It turns out there’s already a dedicated subscription service for this kind of photo culling task, available for just $9.99/month: https://aftershoot.com/
This workflow is purely experimental and built for learning purposes.
