Can you provide your training logs?

[ygtxr1997]

Your training logs can be very helpful for re-implementing SimSwap on other training datasets. We really need these logs for reference. Thank you!

[ygtxr1997]

You said 'Using insightface with different versions is not recommanded. Please use this specific version.' Can you explain more detailedly why? What will happen if I use other versions of face recognition model?

[netrunner-exe]

You said 'Using insightface with different versions is not recommanded. Please use this specific version.' Can you explain more detailedly why? What will happen if I use other versions of face recognition model?

To be honest, it is quite possible to use the latest version of insightface (0.6.2), but for this you need to edit the code a bit. I am not the developer of this repository, I just made an adaptation to work with the latest version of insightface some time ago.

First you need to uninsall old insightface by pip uninstall -y insightface and install latest by pip install insightface==0.6.2 and replace insightface_func/face_detect_crop_multi.py with this one.

# -*- coding: utf-8 -*-
# @Author: netrunner-exe
# @Date:   2022-06-19 18:02:36
# @Last Modified by:   netrunner-exe
# @Last Modified time: 2022-07-29 08:18:31
'''
Author: Naiyuan liu
Github: https://github.com/NNNNAI
Date: 2021-11-23 17:03:58
LastEditors: Naiyuan liu
LastEditTime: 2021-11-24 16:45:41
Description: 
'''
from __future__ import division

import collections
import glob
import os.path as osp

import cv2
import numpy as np
import onnxruntime
from insightface.model_zoo import model_zoo
from insightface.utils import DEFAULT_MP_NAME, ensure_available

from insightface_func.utils import face_align_ffhqandnewarc as face_align

__all__ = ['Face_detect_crop', 'Face']

Face = collections.namedtuple('Face', [
    'bbox', 'kps', 'det_score', 'embedding', 'gender', 'age',
    'embedding_norm', 'normed_embedding',
    'landmark'
])

Face.__new__.__defaults__ = (None, ) * len(Face._fields)

class Face_detect_crop:
    def __init__(self, name=DEFAULT_MP_NAME, root='~/.insightface', allowed_modules=None, **kwargs):
        onnxruntime.set_default_logger_severity(3)
        self.models = {}
        self.model_dir = ensure_available('models', name, root=root)
        onnx_files = glob.glob(osp.join(self.model_dir, '*.onnx'))
        onnx_files = sorted(onnx_files)
        for onnx_file in onnx_files:
            model = model_zoo.get_model(onnx_file, **kwargs)
            if model is None:
                print('model not recognized:', onnx_file)
            elif allowed_modules is not None and model.taskname not in allowed_modules:
                print('model ignore:', onnx_file, model.taskname)
                del model
            elif model.taskname not in self.models and (allowed_modules is None or model.taskname in allowed_modules):
                print('find model:', onnx_file, model.taskname, model.input_shape, model.input_mean, model.input_std)
                self.models[model.taskname] = model
            else:
                print('duplicated model task type, ignore:', onnx_file, model.taskname)
                del model
        assert 'detection' in self.models
        self.det_model = self.models['detection']

    def prepare(self, ctx_id, det_thresh=0.5, det_size=(640, 640), mode ='None'):
        self.det_thresh = det_thresh
        self.mode = mode
        assert det_size is not None
        print('set det-size:', det_size)
        self.det_size = det_size
        for taskname, model in self.models.items():
            if taskname=='detection':
                model.prepare(ctx_id, input_size=det_size, det_thresh=det_thresh)
            else:
                model.prepare(ctx_id)

    def get(self, img, crop_size, max_num=0):
        bboxes, kpss = self.det_model.detect(img,
                                             max_num=max_num,
                                             metric='default')
        if bboxes.shape[0] == 0:
            return None

        ret = []
        align_img_list = []
        M_list = []

        for i in range(bboxes.shape[0]):
            kps = None
            if kpss is not None:
                kps = kpss[i]

            M, _ = face_align.estimate_norm(kps, crop_size, mode = self.mode) 
            align_img = cv2.warpAffine(img, M, (crop_size, crop_size), borderValue=0.0)
            align_img_list.append(align_img)
            M_list.append(M)

        return align_img_list, M_list

Next, you need to replace insightface_func/face_detect_crop_single.py with this one.

# -*- coding: utf-8 -*-
# @Author: netrunner-exe
# @Date:   2022-06-19 18:02:36
# @Last Modified by:   netrunner-exe
# @Last Modified time: 2022-07-29 08:18:24
'''
Author: Naiyuan liu
Github: https://github.com/NNNNAI
Date: 2021-11-23 17:03:58
LastEditors: Naiyuan liu
LastEditTime: 2021-11-24 16:46:04
Description: 
'''
from __future__ import division

import collections
import glob
import os.path as osp

import cv2
import numpy as np
import onnxruntime
from insightface.model_zoo import model_zoo
from insightface.utils import DEFAULT_MP_NAME, ensure_available

from insightface_func.utils import face_align_ffhqandnewarc as face_align

__all__ = ['Face_detect_crop', 'Face']

Face = collections.namedtuple('Face', [
    'bbox', 'kps', 'det_score', 'embedding', 'gender', 'age',
    'embedding_norm', 'normed_embedding',
    'landmark'
])

Face.__new__.__defaults__ = (None, ) * len(Face._fields)

class Face_detect_crop:
    def __init__(self, name=DEFAULT_MP_NAME, root='~/.insightface', allowed_modules=None, **kwargs):
        onnxruntime.set_default_logger_severity(3)
        self.models = {}
        self.model_dir = ensure_available('models', name, root=root)
        onnx_files = glob.glob(osp.join(self.model_dir, '*.onnx'))
        onnx_files = sorted(onnx_files)
        for onnx_file in onnx_files:
            model = model_zoo.get_model(onnx_file, **kwargs)
            if model is None:
                print('model not recognized:', onnx_file)
            elif allowed_modules is not None and model.taskname not in allowed_modules:
                print('model ignore:', onnx_file, model.taskname)
                del model
            elif model.taskname not in self.models and (allowed_modules is None or model.taskname in allowed_modules):
                print('find model:', onnx_file, model.taskname, model.input_shape, model.input_mean, model.input_std)
                self.models[model.taskname] = model
            else:
                print('duplicated model task type, ignore:', onnx_file, model.taskname)
                del model
        assert 'detection' in self.models
        self.det_model = self.models['detection']

    def prepare(self, ctx_id, det_thresh=0.5, det_size=(640, 640), mode ='None'):
        self.det_thresh = det_thresh
        self.mode = mode
        assert det_size is not None
        print('set det-size:', det_size)
        self.det_size = det_size
        for taskname, model in self.models.items():
            if taskname=='detection':
                model.prepare(ctx_id, input_size=det_size, det_thresh=det_thresh)
            else:
                model.prepare(ctx_id)

    def get(self, img, crop_size, max_num=0):
        bboxes, kpss = self.det_model.detect(img,
                                             max_num=max_num,
                                             metric='default')
        if bboxes.shape[0] == 0:
            return None

        det_score = bboxes[..., 4]

        # select the face with the hightest detection score
        best_index = np.argmax(det_score)
        kps = None
        if kpss is not None:
            kps = kpss[best_index]

        M, _ = face_align.estimate_norm(kps, crop_size, mode = self.mode) 
        align_img = cv2.warpAffine(img, M, (crop_size, crop_size), borderValue=0.0)

        return [align_img], [M]

And finally, wherever there is app = Face_detect_crop(name='antelope', root='./insightface_func/models')

Replace with this: app = Face_detect_crop(name='antelope', root='./insightface_func', allowed_modules=['detection'], providers=['CUDAExecutionProvider', 'CPUExecutionProvider'])

It also makes it possible to use antelopev2, buffalo_l, etc. instead of antelope. in Face_detect_crop. But accordingly, you need to download them and unzip them into the ./insightface_func folder.

Or you can try to just change the antelope to what you need and, in theory, the model itself should download. But now it looks like they have problems with the server and therefore you will need to download and unpack it yourself.

[netrunner-exe]

Looks like the automatic download for buffalo_l is working now!)

[netrunner-exe]

By the way, I noticed that buffalo_l and buffalo_sc give a slightly different output.

buffalo_l (result almost same as default antelope):

and buffalo_sc (very lightweight version of buffalo_l, 16MB vs 326MB of buffalo_l):