update app.py

app.py

@@ -4,6 +4,7 @@ warnings.filterwarnings('ignore')
 
 import subprocess, io, os, sys, time
 os.system("pip install gradio==3.36.1")
+import gradio as gr
 
 from loguru import logger
 
@@ -21,8 +22,6 @@ if os.environ.get('IS_MY_DEBUG') is None:
 
 sys.path.insert(0, './GroundingDINO')
 
-import gradio as gr
-
 import argparse
 import copy
 
@@ -301,7 +300,7 @@ def load_lama_cleaner_model():
             device='cpu', # device,
         )
 
-def lama_cleaner_process(image, mask):
+def lama_cleaner_process(image, mask, cleaner_size_limit=1080):
     ori_image = image
     if mask.shape[0] == image.shape[1] and mask.shape[1] == image.shape[0] and mask.shape[0] != mask.shape[1]:
         # rotate image
@@ -311,8 +310,8 @@ def lama_cleaner_process(image, mask):
     original_shape = ori_image.shape
     interpolation = cv2.INTER_CUBIC
     
-    size_limit = 1080
-    if size_limit == "Original":
+    size_limit = cleaner_size_limit
+    if size_limit == -1:
         size_limit = max(image.shape)
     else:
         size_limit = int(size_limit)
@@ -517,7 +516,7 @@ mask_source_draw = "draw a mask on input image"
 mask_source_segment = "type what to detect below"
 
 def run_anything_task(input_image, text_prompt, task_type, inpaint_prompt, box_threshold, text_threshold, 
-            iou_threshold, inpaint_mode, mask_source_radio, remove_mode, remove_mask_extend, num_relation):
+            iou_threshold, inpaint_mode, mask_source_radio, remove_mode, remove_mask_extend, num_relation, cleaner_size_limit=1080):
     if (task_type == 'relate anything'):
         output_images = relate_anything(input_image['image'], num_relation)
         return output_images, gr.Gallery.update(label='relate images')
@@ -644,6 +643,7 @@ def run_anything_task(input_image, text_prompt, task_type, inpaint_prompt, box_t
             image_inpainting = sd_pipe(prompt=inpaint_prompt, image=image_source_for_inpaint, mask_image=image_mask_for_inpaint).images[0]
         else:
             # remove from mask
+            logger.info(f'run_anything_task_[{file_temp}]_{task_type}_5_')
             if mask_source_radio == mask_source_segment:
                 mask_imgs = []
                 masks_shape = masks_ori.shape        
@@ -673,11 +673,16 @@ def run_anything_task(input_image, text_prompt, task_type, inpaint_prompt, box_t
                                         extend_pixels=remove_mask_extend, useRectangle=useRectangle)
                         mask_imgs.append(mask_pil_exp)
                 mask_pil = mix_masks(mask_imgs)
-                output_images.append(mask_pil.convert("RGB"))               
-            image_inpainting = lama_cleaner_process(np.array(image_pil), np.array(mask_pil.convert("L")))
+                output_images.append(mask_pil.convert("RGB"))   
 
+            logger.info(f'run_anything_task_[{file_temp}]_{task_type}_6_')            
+            image_inpainting = lama_cleaner_process(np.array(image_pil), np.array(mask_pil.convert("L")), cleaner_size_limit)
+            output_images.append(image_inpainting)
+
+        logger.info(f'run_anything_task_[{file_temp}]_{task_type}_7_')
         image_inpainting = image_inpainting.resize((image_pil.size[0], image_pil.size[1]))
         output_images.append(image_inpainting)
+        logger.info(f'run_anything_task_[{file_temp}]_{task_type}_9_')
         return output_images, gr.Gallery.update(label='result images')        
     else:
         logger.info(f"task_type:{task_type} error!")

app_cli.py

@@ -115,7 +115,15 @@ if __name__ == '__main__':
                         mask_source_radio = "type what to detect below", 
                         remove_mode = "rectangle",   # ["segment", "rectangle"]
                         remove_mask_extend = "10", 
-                        num_relation = 5)
+                        num_relation = 5,
+                        cleaner_size_limit = -1,
+                        )
     if len(output_images) > 0:
-        logger.info(f'save result to {args.output_image} ... ')
+        logger.info(f'save result to {args.output_image} ... ')        
         output_images[-1].save(args.output_image)
+        # count = 0
+        # for output_image in output_images:
+        #     count += 1
+        #     if isinstance(output_image, np.ndarray):
+        #         output_image = PIL.Image.fromarray(output_image.astype(np.uint8))
+        #     output_image.save(args.output_image.replace(".",  f"_{count}."))