Spaces:
Running
on
A10G
Running
on
A10G
# TODO required tensorflow==2.14 for me | |
# weights from https://github.com/LAION-AI/safety-pipeline/tree/main | |
from PIL import Image | |
import tensorflow_hub as hub | |
import tensorflow | |
import numpy as np | |
import sys | |
sys.path.append('/home/ryn_mote/Misc/generative_recommender/gradio_video/automl/efficientnetv2/') | |
import tensorflow as tf | |
from tensorflow.keras import mixed_precision | |
physical_devices = tf.config.list_physical_devices('GPU') | |
tf.config.experimental.set_memory_growth( | |
physical_devices[0], True | |
) | |
model = tf.keras.models.load_model('nsfweffnetv2-b02-3epochs.h5',custom_objects={"KerasLayer":hub.KerasLayer}) | |
# "The image classifier had been trained on 682550 images from the 5 classes "Drawing" (39026), "Hentai" (28134), "Neutral" (369507), "Porn" (207969) & "Sexy" (37914). | |
# ... we created a manually inspected test set that consists of 4900 samples, that contains images & their captions." | |
# Run prediction | |
def maybe_nsfw(pil_image): | |
# Run prediction | |
imm = tensorflow.image.resize(np.array(pil_image)[:, :, :3], (260, 260)) | |
imm = (imm / 255) | |
pred = model(tensorflow.expand_dims(imm, 0)).numpy() | |
probs = tensorflow.math.softmax(pred[0]).numpy() | |
print(probs) | |
if all([i < .3 for i in probs[[1, 3, 4]]]): | |
return False | |
return True | |
# pre-initializing prediction | |
maybe_nsfw(Image. new("RGB", (260, 260), 255)) | |
model.load_weights('nsfweffnetv2-b02-3epochs.h5', by_name=True, ) | |