Mirrors/modelscope
1
0
Fork 0
mirror of https://github.com/modelscope/modelscope.git synced 2026-02-24 20:19:51 +01:00
Code Issues Packages Projects Releases Wiki Activity

[to #42322933] fix init issues for multi-style cartoon models

Browse Source 
1. 修复多风格模型pipeline初始化问题
        Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10249429
This commit is contained in:
myf272609
2022-09-26 11:55:06 +08:00
committed by Yingda Chen
parent 6e8df4a495
commit ba95cd2460
1 changed files with 7 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
modelscope/pipelines/cv/image_cartoon_pipeline.py
View File

@@ -39,10 +39,13 @@ class ImageCartoonPipeline(Pipeline):
super().__init__(model=model, **kwargs)
with device_placement(self.framework, self.device_name):
self.facer = FaceAna(self.model)
self.sess_anime_head = self.load_sess(
os.path.join(self.model, 'cartoon_h.pb'), 'model_anime_head')
self.sess_anime_bg = self.load_sess(
os.path.join(self.model, 'cartoon_bg.pb'), 'model_anime_bg')
with tf.Graph().as_default():
self.sess_anime_head = self.load_sess(
os.path.join(self.model, 'cartoon_h.pb'),
'model_anime_head')
self.sess_anime_bg = self.load_sess(
os.path.join(self.model, 'cartoon_bg.pb'),
'model_anime_bg')
self.box_width = 288
global_mask = cv2.imread(os.path.join(self.model, 'alpha.jpg'))