Closed pjbh closed 2 years ago
what do you mean exactly? can you share your program and output sample?
DeepFace.verify(filelist[0], filelist[1], model_name = "Facenet")
1/1 [==============================] - 1s 887ms/step 1/1 [==============================] - 0s 39ms/step
It used not to give this output
You can use this arg: https://github.com/serengil/deepface/blob/master/deepface/DeepFace.py#L95
Like this? Doesn't seem to have any effect
for file in filelist[1:len(filelist)]: result = DeepFace.verify(filelist[0], file, model_name = "ArcFace", prog_bar = False) print(file + ' ' + str(result["verified"]) + ' ' + str(result["distance"]))
1/1 [==============================] - 0s 63ms/step 1/1 [==============================] - 0s 56ms/step C:/Users/pjbh2/anaconda3/envs/deepface1\AnnaFriel0001.jpg True 0.376147657150524
interesting because it should work.
what is the content of file list?
Just a few face jpg files, to test things out, so I can see what happens with different models and detectors. It's no biggie for me, just clutters up the output and used not to happen. I still have the old environment if I don't want the new models (Facenet512 appears to give identical results to Facenet, anyhow). To try this version, I set up a fresh env in Anaconda, using Python 3.9.13
if you pass just an image to verify function, then progress bar is silent. i think your trouble is coming from a dependency such as tqdm version. i am closing this issue because i cannot re-produce it.
i had this issue in tensorflow 2.10.0 and i could not resolve it
when i downgrade tensorflow to 2.7.4, then this is sorted.
this is weird?, I'm on tensorflow==2.15.0.post1
this log are still exist
I found the solution
import keras
keras.utils.disable_interactive_logging()
anywhere on the python project (yours or the mtcnn lib) https://github.com/ipazc/mtcnn/issues/121#issuecomment-1942708344
Thank you so much for this amazing piece of work. I downloaded the latest version to get the newly added models. Now the face detectors give an output, a lot of it in the case of mtcnn, which they did not do before. Is there a way to tell them not to? Thanks, Peter