fix mpr and doc (#550)

docs/en_US/tutorials/face_enhancement.md

@@ -18,6 +18,8 @@ faceenhancer = FaceEnhancement()
 img = faceenhancer.enhance_from_image(img)
 ```
 
+Note: please convert the image to float type, currently does not support int8 type.
+
 ### Train (TODO)
 
 In the future, training scripts will be added to facilitate users to train more types of GPEN.

docs/zh_CN/tutorials/face_enhancement.md

@@ -18,6 +18,8 @@ faceenhancer = FaceEnhancement()
 img = faceenhancer.enhance_from_image(img)
 ```
 
+注意：请将图片转为float类型输入，目前不支持int8类型
+
 ### 训练(TODO)
 
 未来还将添加训练脚本方便用户训练出更多类型的 GPEN 人脸增强。

ppgan/apps/mpr_predictor.py

@@ -110,7 +110,7 @@ class MPRPredictor(BasePredictor):
         os.makedirs(self.output_path, exist_ok=True)
         task_path = os.path.join(self.output_path, self.task)
         os.makedirs(task_path, exist_ok=True)
-        image_files = self.get_images(self.images_path)
+        image_files = self.get_images(images_path)
         for image_file in tqdm(image_files):
             img = self.read_image(image_file)
             image_name = os.path.basename(image_file)