NudeNet - automatic censoring AI software

[Miss_Leah.]

Hey guys!

I was wondering if anyone had any experiences with NudeNet. I was hoping to use it to streamline the progress for a current project, but I'm not too skilled yet, currently wrapping up my first year studying computer science. If anyone could share they're knowledge/take a quick look at it to see if they can figure it out, I would love to chat

https://github.com/notAI-tech/NudeNet

Love, Leah

[kerkersklave]

Hi Leah,

it quickly took a look at it as it seemed quite interesting.

I tried it on one random picture and it seems to work very well. I used the installation via pip which at least on Linux worked like a charm. What is the project you're working on? And how can I help you?

[Miss_Leah.]

Hi Leah,

it quickly took a look at it as it seemed quite interesting.

I tried it on one random picture and it seems to work very well. I used the installation via pip which at least on Linux worked like a charm. What is the project you're working on? And how can I help you?

Thanks for taking a look at it :)

Do you mind sending the files from the folder you used to run it? I get the detector to work and analyze the photos, but when I run the censor method I get an error say the "modelproto does not have a graph" or something like that. Googling it presents me with all sorts of nonsense I don't understand, so I'm at a loss here.

Im currently fiddling with a tease ai personality, and was hoping to do a bit of censoring in there, not sure how its gonna work, but either way it seems like its a great tool, that can censor videos as well, as far as I understand

[kerkersklave]

I've tested it on the following picture. It censors face, breasts, and pussy. I've tried other pictures but those are private


Which API are you using? The python one as well? I have not seen that error message.

[bonsec]

I used this as well for a project, however I only used their detect method and did my own censoring in frontend javascript.
I had a quick look at the error and it comes from the detector model, so if detect is working fine but censor is giving that error, I suspect the args passed to censor had some mixup.

[Miss_Leah.]

I've tested it on the following picture. It censors face, breasts, and pussy. I've tried other pictures but those are private


Which API are you using? The python one as well? I have not seen that error message.

I was actually mainly referring to the python file, I'm unsure if I'm writing the code right, I haven't really gotten the hang of interpreting other peoples code, and its been a short while since I dabbled with python last, do you mind sending it?

[Miss_Leah.]

I used this as well for a project, however I only used their detect method and did my own censoring in frontend javascript.
I had a quick look at the error and it comes from the detector model, so if detect is working fine but censor is giving that error, I suspect the args passed to censor had some mixup.

I'm passing the same args as with the detect method, so it shouldn't be any issue, but maybe im wrong, literally copied the code from a guide, and changed the filenames to match. The guide in question is the following

https://ourcodeworld.com/articles/read/ ... -in-python

[kerkersklave]

The guide matches what I did. Can you maybe copy a minimal code example that causes the error? Best just copy it, sometimes one can read their own code over and over again without spotting a subtle problem or difference.

[kerkersklave]

I was actually mainly referring to the python file, I'm unsure if I'm writing the code right, I haven't really gotten the hang of interpreting other peoples code, and its been a short while since I dabbled with python last, do you mind sending it?

My code looks like this:

Code: Select all

from nudenet import NudeDetector

detector = NudeDetector()
detector.censor('test.jpeg', out_path = 'out.jpeg')

[Miss_Leah.]

I was actually mainly referring to the python file, I'm unsure if I'm writing the code right, I haven't really gotten the hang of interpreting other peoples code, and its been a short while since I dabbled with python last, do you mind sending it?

My code looks like this:

Code: Select all

from nudenet import NudeDetector

detector = NudeDetector()
detector.censor('test.jpeg', out_path = 'out.jpeg')

My code looks like the following

Code: Select all

from nudenet import NudeDetector
from nudenet import NudeClassifier

detector = NudeDetector()

detector.censor(
    './image1.jpeg', 
    out_path='./image1_censored.jpeg', 
    visualize=False
)

[Miss_Leah.]

Copying your code didn't seem to do anything for me either.. /:

[kerkersklave]

Your code works fine for me, can't see any problem either.

This error comes somewhere from within the neural net engine from googling it. Sounds like the model has not been loaded properly. From looking at the code of censor() it just calls detect() and then manipulates the images.

Did you try calling detect() directly and does that work? (Not quite sure form your previous posts).
Can you copy the complete error message including the stack trace?

[Miss_Leah.]

Your code works fine for me, can't see any problem either.

This error comes somewhere from within the neural net engine from googling it. Sounds like the model has not been loaded properly. From looking at the code of censor() it just calls detect() and then manipulates the images.

Did you try calling detect() directly and does that work? (Not quite sure form your previous posts).
Can you copy the complete error message including the stack trace?

Aha, detect() throws the same error, here's the complete error message

Code: Select all

Traceback (most recent call last):
  File "test.py", line 3, in <module>
    detector = NudeDetector()
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/nudenet/detector.py", line 57, in __init__
    self.detection_model = onnxruntime.InferenceSession(checkpoint_path)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 280, in __init__
    self._create_inference_session(providers, provider_options)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 307, in _create_inference_session
    sess = C.InferenceSession(session_options, self._model_path, True, self._read_config_from_model)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from /Users/leah/.NudeNet/detector_v2_default_checkpoint.onnx failed:/Users/runner/work/1/s/onnxruntime/core/graph/model.cc:101 onnxruntime::Model::Model(onnx::ModelProto &&, const onnxruntime::PathString &, const onnxruntime::IOnnxRuntimeOpSchemaRegistryList *, const logging::Logger &) ModelProto does not have a graph.

[Miss_Leah.]

Aha, even my noob ass can see that its the constructor that throws the error now, not that it explains much to me

[kerkersklave]

When I first ran the detector, it did download the model file and put it in

Code: Select all

.NudeNet/detector_v2_default_checkpoint.onnx

in my home directory.

Does that file exist? Should be about 140MB.