rockyzhengwu / nsfw Goto Github PK

View Code? Open in Web Editor NEW

550.0 550.0 209.0 2.83 MB

Not Suitable for Work (NSFW) classification

License: Apache License 2.0

Python 98.85% Shell 1.15%

nsfw's Introduction

NSFW Model

This repo contains code for running Not Suitable for Work (NSFW) classification.

online demo

Usage

script

python nsfw_predict.py /tmp/test/test.jpeg

result :

{'class': 'sexy', 'probability': {'drawings': 0.008320281, 'hentai': 0.0011919827, 'neutral': 0.13077603, 'porn': 0.13146976, 'sexy': 0.72824186}}

can find the meaning of every label at repo nsfw_data_scrapper

Deploy by TensorFlow Serving

your have to install Tensorflow Serving first

start the server

./start_tensorflow_serving.sh

test server

python serving_client.py /tmp/test/test.jpeg

Train

train code at resnet

train a new model

convert source to tfrecord user convert_image_to_tfrecord.py
train a model from scratch or fine tune

the model code copy from Tensorflow offical model

Data

you can find the detail at repo nsfw_data_scrapper

nsfw's People

Contributors

Stargazers

Watchers

Forkers

skymysky justintung 496080199 xiaomin0322 dulumao b-xiang alexshilucky ligo zhsker twxjlmq xuningnb guyueyuqi hellomeme iamxiaoma davidalphafox fitear awesome-ml schr0dingercat guozhanxian netww lgb020 xboge wikipedia2008 jdc08161063 wangkanger davidce nilportugues yushu-liu apple006 lanlanx cn-tim kufan dayinruoxi jellychendeveloper ruffe-ai beanecessar kylinliu hanjack abccyz btbujiangjun heyang5254 jarvis-su garxin mickeyto nipengmath jerryjia yenwudr cassdsghxw xj52hp zvrr isqxie sp0917 942bc granbluefu hustlemickey accuser321 xmmwc keyman9848 wowliao jinshengzhang jackdsunshine famoustang harveywei nottellyou chengchuanglin zzhuuh2 dotawtf iceelor vaguezhao timovip wangjie147 wuyeguo rokerdou creatchange hihiworld eddylapis qwer4301 yongchengzhao cwk8866 meisen2008 starsw001 vb0067 antediluvian rs-v bigdumpling objects-time shuzhili jerryliu306 wuyuan0717 thdongl codenewer fengjinli iaep ystar2016 xcorg evifree tthhee cndavy suyangshuo fengkuangfj

nsfw's Issues

finetune

finetune的是哪个模型呀……我试了resnet预训练好的几个模型，都结构对不上

File "nsfw_predict.py", line 42, in predict
feed_dict={inputs: [image_data] * _BATCH_SIZE})
File "\nsfw-master\venv\lib\site-packages\tensorflow\python\client\session.py", line 950, in run
run_metadata_ptr)
File "\nsfw-master\venv\lib\site-packages\tensorflow\python\client\session.py", line 1149, in _run
str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (128, 64, 64) for Tensor 'input_tensor:0', which has shape '(128, 64, 64, 3)'

tensorflow2.0 使用需要更改兼容方法 tf.compat.v1

are you kidding me?

上传了我儿子图片结果竟然这样的

Could you plz provide the tfrecords file? Thanks a lot !

How to predict using ckpt files?

Hi! I have trained a new model via running nsfw_main_finetune.py and got some ckpt files. But how may I use ckpt files to predict a test image without converting them to pb files? thanks a lot !

Convert this model to tflite

When i try to convert this model to tflite , i get this error. any idea how to fix this ?

b'/bin/sh: toco_from_protos: command not found\n' None

Code: converter = lite.TFLiteConverter.from_saved_model(_MODEL_DIR) tflite_model = converter.convert()

为什么预测同一张图片，分数不同？

使用http://ai.midday.me 鉴别yellow图功能检测出来的分数，和手动运行python nsfw_predict.py + 路径得到的分数不一样，有时甚至差很多，以至于分类不同。
是模型不一样么？感觉线上的检测效果会更好。
如果模型不一样，能不能提供一下线上的模型呀？感谢！

On line test is brilliant, but is there any way to improve the model performance?

I test the model with a few pitcures which is the close up of the sexual organs. The model fails to classify them and concludes that these pitcures are suitable for work.