DeepFace Based Batch Face Similarity Sorting Gradio APP For Windows, RunPod and Massed Compute - 1-Click to Install - Uses TensorFlow GPU - Very Fast

Hockey

I managed to type in the model name for most of them. However there are two models that won't work that way. Dlib and Human-beings. Dlib I managed to install into your environment, however Human-beings I can't get to work... Since they are up there with Facenet in accuracy, it would be great if we could use them. There also seem to be a new version out since it has a model not available in v9. https://github.com/serengil/deepface/blob/master/deepface/models/facial_recognition/Buffalo_L.py Would be great if we could use that one too. I'm testing all the models extensively. Also have you heard about kbyAI? https://github.com/kby-ai/FaceRecognition-Windows/tree/main/lib

Hockey

Is it possible to add the other models as well? Human-beings Facenet Dlib VGG-Face ArcFace GhostFaceNet SFace OpenFace DeepFace DeepID

Hockey

Thank you, works with v9. =)

Hockey

Furkan Gözükara

good idea let me add that as optional

Hockey

Is there a way to not get the average values but the best match value to a source image? Some of the source images are from a view where the face is hardly visible and that will bring down the average value. Here's an example from another software I use: Compared with IMG_20190306_183635.png: 45.90% Compared with IMG_20191011_171158.png: 56.40% Compared with IMG_20191011_171202.png: 0.00% Compared with IMG_20191011_171205.png: 36.24% Best match: DSC_0536.png with 62.42% It checks against each of the source images and presents a value, at the end it also says the average match value and the best individual match value.

Furkan Gözükara

did you try to run? i think it should work . but tensorflow is just a very bad library sadly

Hockey

Got Windows 11 and use Nvidia RTX 3080 16 Gb on a laptop. I get this error when starting the software: 2025-06-18 13:59:37.877925: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found 2025-06-18 13:59:37.878240: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine. During install I got these errors: WARNING: Ignoring invalid distribution -ip (c:\python310\lib\site-packages) Requirement already satisfied: pip in c:\python310\lib\site-packages (25.1.1) WARNING: Ignoring invalid distribution -ip (c:\python310\lib\site-packages) WARNING: Ignoring invalid distribution -ip (c:\python310\lib\site-packages) ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. tensorflow-gpu 2.10.0 requires keras<2.11,>=2.10.0, but you have keras 3.10.0 which is incompatible. tensorflow-gpu 2.10.0 requires protobuf<3.20,>=3.9.2, but you have protobuf 5.29.5 which is incompatible. tensorflow-gpu 2.10.0 requires tensorboard<2.11,>=2.10, but you have tensorboard 2.19.0 which is incompatible. I have Cuda 12.8 and 12.4 installed, tried to move Cuda 12.4 up in the path list ahead of 12.8 but did not help. Got Cuddn 8.9.7 and 9.7 installed. I don't have any issues with your other softwares. The software starts and start working but after fixing the images nothing happens.

Furkan Gözükara

this can be used to sort images based on similarity to certain given images

Markus

For what do we use that?

Furkan Gözükara

added that feature with v6

Lou

Will this also include sub folders if needed?