qhgz2013 / anime-face-detector Goto Github PK

View Code? Open in Web Editor NEW

240.0 7.0 32.0 4.37 MB

A Faster-RCNN based anime face detector implementation using tensorflow.

License: MIT License

Python 96.62% Makefile 0.04% Batchfile 0.08% C++ 0.06% Cuda 1.93% Cython 1.28%

faster-rcnn face-detection anime

anime-face-detector's Introduction

Anime-Face-Detector

A Faster-RCNN based anime face detector.

This detector is trained on 6000 training samples and 641 testing samples, randomly selected from the dataset which is crawled from top 100 pixiv daily ranking.

Thanks to OpenCV based Anime face detector written by nagadomi, which helps labelling the data.

The original implementation of Faster-RCNN using Tensorflow can be found here

Dependencies

Python >= 3.6
tensorflow latest 1.x or 2.x
opencv-python (Will use other packages like pillow and scikit-image as backend in future version)
cython (optional, can be ignored with additional -nms-type PY_NMS argument)
Pre-trained ResNet101 model

Usage

Clone this repository

git clone https://github.com/qhgz2013/anime-face-detector.git

Download the pre-trained model
Google Drive: here
Baidu Netdisk: here
Unzip the model file into model directory
Build the CPU NMS model (skip this step if use PY_NMS with argument: -nms-type PY_NMS)
```
make clean
make
```
If using Windows Power Shell, type cmd /C make.bat to run build script.

Run the demo as you want

Visualize the result (without output path):
```
python main.py -i /path/to/image.jpg
```

Save results to a json file

python main.py -i /path/to/image.jpg -o /path/to/output.json

Format: {"image_path": [{"score": predicted_probability, "bbox": [min_x, min_y, max_x, max_y]}, ...], ...} Sample output file:

{"/path/to/image.jpg": [{"score": 0.9999708, "bbox": [551.3375, 314.50253, 729.2599, 485.25674]}]}

Detecting a whole directory with recursion

python main.py -i /path/to/dir -o /path/to/output.json

Customize threshold

python main.py -i /path/to/image.jpg -nms 0.3 -conf 0.8

Customize model path

python main.py -i /path/to/image.jpg -model /path/to/model.ckpt

Customize nms type (supports CPU_NMS and PY_NMS, not supports GPU_NMS because of the complicated build process for Windows platform)
```
python main.py -i /path/to/image.jpg -nms-type PY_NMS
```
Crop detected images and store them in a folder (start output is an integer to start naming the cropped images, default is 0)
```
python main.py -i /path/to/image/or/folder -crop-location /path/to/store/cropped/images -start-output 1
```

Crop detected images and resizes them

python main.py -i /path/to/image/or/folder -crop-location /path/to/store/cropped/images -crop-height 224 -crop-width 224

Results

Mean AP for this model: 0.9086

About training

This model is directly trained by Faster-RCNN, with following argument:

python tools/trainval_net.py --weight data/imagenet_weights/res101.ckpt --imdb voc_2007_trainval --imdbval voc_2007_test --iters 60000 --cfg experiments/cfgs/res101.yml --net res101 --set ANCHOR_SCALES "[4,8,16,32]" ANCHOR_RATIOS "[1]" TRAIN.STEPSIZE "[50000]"

Dataset

We've uploaded the dataset to Google drive here, dataset structure is similar to VOC2007 (used in original Faster-RCNN implementation).

Citation and declaration

Feel free to cite this repo and dataset.
This work is not related to my research team and lab, just my personal interest.

anime-face-detector's People

Contributors

Stargazers

Watchers

anime-face-detector's Issues

George Costanza is anime?

This has worked super well, except it has a very weird outlier of recognizing George Costanza's bald head as anime. I've tested this with 3 images of him:

I'm not totally sure why this is so set on believing George Costanza is anime, but maybe it is right?!?! 🤔

No module named 'tensorflow.contrib'

I think this means you're still using tensorflow pre-2.0, how much trouble would it be to upgrade? I'll make a PR if you say it won't be too hard

Cant build model

After !python setup.py build_ext --inplace
Get roor:
TypeError: can only concatenate list (not "dict") to list

Full code
https://colab.research.google.com/drive/1wLox24qohgrGiCr90gdnwRzQLZ54LGHY

Is there a way to get the location of the eyes?

error: Unable to find vcvarsall.bat

Hi everyone. I try to run the command "make". But I see the error: Unable to find vcvarsall.bat.

Ps: I run it on Windows with anaconda
Thanks

tensorflow.python.framework.errors_impl.NotFoundError: model; No such file or directory

Hello! When i run the command below, it can not work. Can someone help me?

python /home/beryl/Documents/anime-face-detector-master/main.py -i /home/beryl/Documents/Anime-Face-Dataset-master/animeface-character-dataset/thumb/000_hatsune_miku/ -o /home/beryl/Documents/character_dataset/000_hatsune_miku.json -nms-type PY_NMS

/home/beryl/Documents/Anime-Face-Dataset-master/animeface-character-dataset/thumb/000_hatsune_miku/
2019-11-28 20:33:46.663551: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-11-28 20:33:47.276064: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:897] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-11-28 20:33:47.276415: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1405] Found device 0 with properties:
name: GeForce RTX 2070 major: 7 minor: 5 memoryClockRate(GHz): 1.71
pciBusID: 0000:01:00.0
totalMemory: 7.79GiB freeMemory: 7.69GiB
2019-11-28 20:33:47.276443: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1484] Adding visible gpu devices: 0
2019-11-28 20:33:47.537843: I tensorflow/core/common_runtime/gpu/gpu_device.cc:965] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-11-28 20:33:47.537891: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971] 0
2019-11-28 20:33:47.537897: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] 0: N
2019-11-28 20:33:47.538050: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1097] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7402 MB memory) -> physical GPU (device: 0, name: GeForce RTX 2070, pci bus id: 0000:01:00.0, compute capability: 7.5)
Traceback (most recent call last):
File "/home/beryl/Documents/anime-face-detector-master/main.py", line 172, in
main()
File "/home/beryl/Documents/anime-face-detector-master/main.py", line 124, in main
saver.restore(sess, args.model)
File "/home/beryl/anaconda3/envs/py3.6/lib/python3.6/site-packages/tensorflow/python/training/saver.py", line 1715, in restore
if not checkpoint_exists(compat.as_text(save_path)):
File "/home/beryl/anaconda3/envs/py3.6/lib/python3.6/site-packages/tensorflow/python/training/saver.py", line 2056, in checkpoint_exists
if file_io.get_matching_files(pathname):
File "/home/beryl/anaconda3/envs/py3.6/lib/python3.6/site-packages/tensorflow/python/lib/io/file_io.py", line 342, in get_matching_files
for single_filename in filename
File "/home/beryl/anaconda3/envs/py3.6/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 519, in exit
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.NotFoundError: model; No such file or directory

Can I access to the training data?

Hi! Thank you for creating great model.

Is it possible to access to the training data you used for this model?
Guessing from the README, you prepared rough annotated data and adjusted the data manually. Is it right? It's ok to make such data by myself but it's very helpful if I can use existing data.

Japanese filenames cause an error

$ python main.py -i ~/Desktop/a/ -crop-location ~/Desktop/b/

[1/144] Elapsed: 0:00.000, ETA: 0:00.000 >> C:/Users/me/Desktop/a/censored_cg05a.png
[2/144] Elapsed: 0:03.402, ETA: 8:06.521 >> C:/Users/me/Desktop/a/censored_cg05b.png
[3/144] Elapsed: 0:05.996, ETA: 7:05.748 >> C:/Users/me/Desktop/a/censored_cg05c.png
[4/144] Elapsed: 0:08.626, ETA: 6:45.443 >> C:/Users/me/Desktop/a/censored_cg05d.png
[5/144] Elapsed: 0:11.273, ETA: 6:34.558 >> C:/Users/me/Desktop/a/censored_cg14a.png
[6/144] Elapsed: 0:13.825, ETA: 6:24.344 >> C:/Users/me/Desktop/a/censored_ct07a.png
[7/144] Elapsed: 0:16.278, ETA: 6:14.415 >> C:/Users/me/Desktop/a/cg08ax2.png
Traceback (most recent call last):
  File "main.py", line 187, in <module>
    main()
  File "main.py", line 138, in main
    print('[%d/%d] Elapsed: %s, ETA: %s >> %s' % (idx+1, file_len, fmt_time(elapsed), fmt_time(eta), file))
  File "C:\Users\me\AppData\Local\Programs\Python\Python38\lib\encodings\cp1252.py", line 19, in encode
    return codecs.charmap_encode(input,self.errors,encoding_table)[0]
UnicodeEncodeError: 'charmap' codec can't encode characters in position 70-72: character maps to <undefined>
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile

It's failing on this

If I rename them with only ascii, it works.