Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -71,18 +71,18 @@ if not os.path.exists(reference_model_path):
|
|
71 |
reference_detector_global = YOLO(reference_model_path)
|
72 |
print("YOLO reference model loaded in {:.2f} seconds".format(time.time() - start_time))
|
73 |
|
74 |
-
print("Loading U²-Net model for reference background removal (U2NETP)...")
|
75 |
-
start_time = time.time()
|
76 |
-
u2net_model_path = os.path.join(CACHE_DIR, "u2netp.pth")
|
77 |
-
if not os.path.exists(u2net_model_path):
|
78 |
-
|
79 |
-
|
80 |
-
u2net_global = U2NETP(3, 1)
|
81 |
-
u2net_global.load_state_dict(torch.load(u2net_model_path, map_location="cpu"))
|
82 |
-
device = "cpu"
|
83 |
-
u2net_global.to(device)
|
84 |
-
u2net_global.eval()
|
85 |
-
print("U²-Net model loaded in {:.2f} seconds".format(time.time() - start_time))
|
86 |
|
87 |
print("Loading BiRefNet model...")
|
88 |
start_time = time.time()
|
@@ -119,16 +119,16 @@ def unload_and_reload_models():
|
|
119 |
new_birefnet = AutoModelForImageSegmentation.from_pretrained(
|
120 |
"zhengpeng7/BiRefNet", trust_remote_code=True, cache_dir=CACHE_DIR
|
121 |
)
|
122 |
-
new_birefnet.to(device)
|
123 |
-
new_birefnet.eval()
|
124 |
-
new_u2net = U2NETP(3, 1)
|
125 |
-
new_u2net.load_state_dict(torch.load(os.path.join(CACHE_DIR, "u2netp.pth"), map_location="cpu"))
|
126 |
-
new_u2net.to(device)
|
127 |
-
new_u2net.eval()
|
128 |
drawer_detector_global = new_drawer_detector
|
129 |
reference_detector_global = new_reference_detector
|
130 |
birefnet_global = new_birefnet
|
131 |
-
u2net_global =
|
132 |
print("Models reloaded in {:.2f} seconds".format(time.time() - start_time))
|
133 |
|
134 |
# ---------------------
|
@@ -159,23 +159,27 @@ def detect_reference_square(img: np.ndarray):
|
|
159 |
res[0].cpu().boxes.xyxy[0]
|
160 |
)
|
161 |
|
162 |
-
#
|
163 |
-
def
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
|
168 |
-
|
169 |
-
|
170 |
-
|
171 |
-
|
172 |
-
|
173 |
-
|
174 |
-
|
175 |
-
|
176 |
-
|
177 |
-
|
178 |
-
#
|
|
|
|
|
|
|
|
|
179 |
def remove_bg(image: np.ndarray) -> np.ndarray:
|
180 |
t = time.time()
|
181 |
image_pil = Image.fromarray(image)
|
@@ -187,7 +191,7 @@ def remove_bg(image: np.ndarray) -> np.ndarray:
|
|
187 |
scale_ratio = 1024 / max(image_pil.size)
|
188 |
scaled_size = (int(image_pil.size[0] * scale_ratio), int(image_pil.size[1] * scale_ratio))
|
189 |
result = np.array(pred_pil.resize(scaled_size))
|
190 |
-
print("BiRefNet
|
191 |
return result
|
192 |
|
193 |
def make_square(img: np.ndarray):
|
@@ -469,6 +473,7 @@ def predict(
|
|
469 |
print("Drawer detection completed in {:.2f} seconds".format(time.time() - t))
|
470 |
except DrawerNotDetectedError as e:
|
471 |
return None, None, None, None, f"Error: {str(e)}"
|
|
|
472 |
t = time.time()
|
473 |
shrunked_img = make_square(shrink_bbox(drawer_img, 0.90))
|
474 |
del drawer_img
|
@@ -490,9 +495,9 @@ def predict(
|
|
490 |
# ---------------------
|
491 |
t = time.time()
|
492 |
reference_obj_img = make_square(reference_obj_img)
|
493 |
-
|
494 |
-
reference_square_mask = remove_bg_reference(reference_obj_img)
|
495 |
print("Reference image processing completed in {:.2f} seconds".format(time.time() - t))
|
|
|
496 |
t = time.time()
|
497 |
try:
|
498 |
cv2.imwrite("mask.jpg", cv2.cvtColor(reference_obj_img, cv2.COLOR_RGB2GRAY))
|
@@ -565,6 +570,7 @@ def predict(
|
|
565 |
del objects_mask
|
566 |
gc.collect()
|
567 |
print("Mask dilation completed in {:.2f} seconds".format(time.time() - t))
|
|
|
568 |
Image.fromarray(dilated_mask).save("./outputs/scaled_mask_new.jpg")
|
569 |
|
570 |
# ---------------------
|
@@ -573,12 +579,16 @@ def predict(
|
|
573 |
t = time.time()
|
574 |
outlines, contours = extract_outlines(dilated_mask)
|
575 |
print("Outline extraction completed in {:.2f} seconds".format(time.time() - t))
|
|
|
576 |
output_img = shrunked_img.copy()
|
577 |
del shrunked_img
|
578 |
gc.collect()
|
|
|
579 |
t = time.time()
|
580 |
use_finger_clearance = True if finger_clearance.lower() == "yes" else False
|
581 |
-
doc, final_polygons_inch = save_dxf_spline(
|
|
|
|
|
582 |
del contours
|
583 |
gc.collect()
|
584 |
print("DXF generation completed in {:.2f} seconds".format(time.time() - t))
|
@@ -623,8 +633,14 @@ def predict(
|
|
623 |
text_x = (inner_min_x + inner_max_x) / 2.0
|
624 |
text_height_dxf = 0.5
|
625 |
text_y_dxf = inner_min_y - 0.125 - text_height_dxf
|
626 |
-
text_entity = msp.add_text(
|
627 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
628 |
text_entity.dxf.insert = (text_x, text_y_dxf)
|
629 |
|
630 |
# Save the DXF
|
@@ -644,8 +660,27 @@ def predict(
|
|
644 |
text_y_in = inner_min_y - 0.125 - text_height_cv
|
645 |
text_y_img = int(processed_size[0] - (text_y_in / scaling_factor))
|
646 |
org = (text_x_img - int(len(annotation_text.strip()) * 6), text_y_img)
|
647 |
-
|
648 |
-
cv2.putText(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
649 |
|
650 |
# Restore brightness for display purposes:
|
651 |
# Since we reduced brightness by 0.5 during preprocessing,
|
@@ -656,11 +691,14 @@ def predict(
|
|
656 |
|
657 |
outlines_color = cv2.cvtColor(new_outlines, cv2.COLOR_BGR2RGB)
|
658 |
print("Total prediction time: {:.2f} seconds".format(time.time() - overall_start))
|
659 |
-
|
660 |
-
|
661 |
-
|
662 |
-
|
663 |
-
|
|
|
|
|
|
|
664 |
|
665 |
# ---------------------
|
666 |
# Gradio Interface
|
|
|
71 |
reference_detector_global = YOLO(reference_model_path)
|
72 |
print("YOLO reference model loaded in {:.2f} seconds".format(time.time() - start_time))
|
73 |
|
74 |
+
# print("Loading U²-Net model for reference background removal (U2NETP)...")
|
75 |
+
# start_time = time.time()
|
76 |
+
# u2net_model_path = os.path.join(CACHE_DIR, "u2netp.pth")
|
77 |
+
# if not os.path.exists(u2net_model_path):
|
78 |
+
# print("Caching U²-Net model to", u2net_model_path)
|
79 |
+
# shutil.copy("u2netp.pth", u2net_model_path)
|
80 |
+
# u2net_global = U2NETP(3, 1)
|
81 |
+
# u2net_global.load_state_dict(torch.load(u2net_model_path, map_location="cpu"))
|
82 |
+
# device = "cpu"
|
83 |
+
# u2net_global.to(device)
|
84 |
+
# u2net_global.eval()
|
85 |
+
# print("U²-Net model loaded in {:.2f} seconds".format(time.time() - start_time))
|
86 |
|
87 |
print("Loading BiRefNet model...")
|
88 |
start_time = time.time()
|
|
|
119 |
new_birefnet = AutoModelForImageSegmentation.from_pretrained(
|
120 |
"zhengpeng7/BiRefNet", trust_remote_code=True, cache_dir=CACHE_DIR
|
121 |
)
|
122 |
+
# new_birefnet.to(device)
|
123 |
+
# new_birefnet.eval()
|
124 |
+
# new_u2net = U2NETP(3, 1)
|
125 |
+
# new_u2net.load_state_dict(torch.load(os.path.join(CACHE_DIR, "u2netp.pth"), map_location="cpu"))
|
126 |
+
# new_u2net.to(device)
|
127 |
+
# new_u2net.eval()
|
128 |
drawer_detector_global = new_drawer_detector
|
129 |
reference_detector_global = new_reference_detector
|
130 |
birefnet_global = new_birefnet
|
131 |
+
u2net_global = new_birefnet
|
132 |
print("Models reloaded in {:.2f} seconds".format(time.time() - start_time))
|
133 |
|
134 |
# ---------------------
|
|
|
159 |
res[0].cpu().boxes.xyxy[0]
|
160 |
)
|
161 |
|
162 |
+
# Use U2NETP for reference background removal.
|
163 |
+
# def remove_bg_u2netp(image: np.ndarray) -> np.ndarray:
|
164 |
+
# t = time.time()
|
165 |
+
# image_pil = Image.fromarray(image)
|
166 |
+
# transform_u2netp = transforms.Compose([
|
167 |
+
# transforms.Resize((320, 320)),
|
168 |
+
# transforms.ToTensor(),
|
169 |
+
# transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
|
170 |
+
# ])
|
171 |
+
# input_tensor = transform_u2netp(image_pil).unsqueeze(0).to("cpu")
|
172 |
+
# with torch.no_grad():
|
173 |
+
# outputs = u2net_global(input_tensor)
|
174 |
+
# pred = outputs[0]
|
175 |
+
# pred = (pred - pred.min()) / (pred.max() - pred.min() + 1e-8)
|
176 |
+
# pred_np = pred.squeeze().cpu().numpy()
|
177 |
+
# pred_np = cv2.resize(pred_np, (image_pil.width, image_pil.height))
|
178 |
+
# pred_np = (pred_np * 255).astype(np.uint8)
|
179 |
+
# print("U2NETP background removal completed in {:.2f} seconds".format(time.time() - t))
|
180 |
+
# return pred_np
|
181 |
+
|
182 |
+
# Use BiRefNet for main object background removal.
|
183 |
def remove_bg(image: np.ndarray) -> np.ndarray:
|
184 |
t = time.time()
|
185 |
image_pil = Image.fromarray(image)
|
|
|
191 |
scale_ratio = 1024 / max(image_pil.size)
|
192 |
scaled_size = (int(image_pil.size[0] * scale_ratio), int(image_pil.size[1] * scale_ratio))
|
193 |
result = np.array(pred_pil.resize(scaled_size))
|
194 |
+
print("BiRefNet background removal completed in {:.2f} seconds".format(time.time() - t))
|
195 |
return result
|
196 |
|
197 |
def make_square(img: np.ndarray):
|
|
|
473 |
print("Drawer detection completed in {:.2f} seconds".format(time.time() - t))
|
474 |
except DrawerNotDetectedError as e:
|
475 |
return None, None, None, None, f"Error: {str(e)}"
|
476 |
+
# Ensure that shrunked_img is defined only after successful detection.
|
477 |
t = time.time()
|
478 |
shrunked_img = make_square(shrink_bbox(drawer_img, 0.90))
|
479 |
del drawer_img
|
|
|
495 |
# ---------------------
|
496 |
t = time.time()
|
497 |
reference_obj_img = make_square(reference_obj_img)
|
498 |
+
reference_square_mask = remove_bg(reference_obj_img)
|
|
|
499 |
print("Reference image processing completed in {:.2f} seconds".format(time.time() - t))
|
500 |
+
|
501 |
t = time.time()
|
502 |
try:
|
503 |
cv2.imwrite("mask.jpg", cv2.cvtColor(reference_obj_img, cv2.COLOR_RGB2GRAY))
|
|
|
570 |
del objects_mask
|
571 |
gc.collect()
|
572 |
print("Mask dilation completed in {:.2f} seconds".format(time.time() - t))
|
573 |
+
|
574 |
Image.fromarray(dilated_mask).save("./outputs/scaled_mask_new.jpg")
|
575 |
|
576 |
# ---------------------
|
|
|
579 |
t = time.time()
|
580 |
outlines, contours = extract_outlines(dilated_mask)
|
581 |
print("Outline extraction completed in {:.2f} seconds".format(time.time() - t))
|
582 |
+
|
583 |
output_img = shrunked_img.copy()
|
584 |
del shrunked_img
|
585 |
gc.collect()
|
586 |
+
|
587 |
t = time.time()
|
588 |
use_finger_clearance = True if finger_clearance.lower() == "yes" else False
|
589 |
+
doc, final_polygons_inch = save_dxf_spline(
|
590 |
+
contours, scaling_factor, processed_size[0], finger_clearance=use_finger_clearance
|
591 |
+
)
|
592 |
del contours
|
593 |
gc.collect()
|
594 |
print("DXF generation completed in {:.2f} seconds".format(time.time() - t))
|
|
|
633 |
text_x = (inner_min_x + inner_max_x) / 2.0
|
634 |
text_height_dxf = 0.5
|
635 |
text_y_dxf = inner_min_y - 0.125 - text_height_dxf
|
636 |
+
text_entity = msp.add_text(
|
637 |
+
annotation_text.strip(),
|
638 |
+
dxfattribs={
|
639 |
+
"height": text_height_dxf,
|
640 |
+
"layer": "ANNOTATION",
|
641 |
+
"style": "Bold"
|
642 |
+
}
|
643 |
+
)
|
644 |
text_entity.dxf.insert = (text_x, text_y_dxf)
|
645 |
|
646 |
# Save the DXF
|
|
|
660 |
text_y_in = inner_min_y - 0.125 - text_height_cv
|
661 |
text_y_img = int(processed_size[0] - (text_y_in / scaling_factor))
|
662 |
org = (text_x_img - int(len(annotation_text.strip()) * 6), text_y_img)
|
663 |
+
|
664 |
+
cv2.putText(
|
665 |
+
output_img,
|
666 |
+
annotation_text.strip(),
|
667 |
+
org,
|
668 |
+
cv2.FONT_HERSHEY_SIMPLEX,
|
669 |
+
1.3,
|
670 |
+
(0, 0, 255),
|
671 |
+
3,
|
672 |
+
cv2.LINE_AA
|
673 |
+
)
|
674 |
+
cv2.putText(
|
675 |
+
new_outlines,
|
676 |
+
annotation_text.strip(),
|
677 |
+
org,
|
678 |
+
cv2.FONT_HERSHEY_SIMPLEX,
|
679 |
+
1.3,
|
680 |
+
(0, 0, 255),
|
681 |
+
3,
|
682 |
+
cv2.LINE_AA
|
683 |
+
)
|
684 |
|
685 |
# Restore brightness for display purposes:
|
686 |
# Since we reduced brightness by 0.5 during preprocessing,
|
|
|
691 |
|
692 |
outlines_color = cv2.cvtColor(new_outlines, cv2.COLOR_BGR2RGB)
|
693 |
print("Total prediction time: {:.2f} seconds".format(time.time() - overall_start))
|
694 |
+
|
695 |
+
return (
|
696 |
+
cv2.cvtColor(output_img, cv2.COLOR_BGR2RGB),
|
697 |
+
outlines_color,
|
698 |
+
dxf_filepath,
|
699 |
+
dilated_mask,
|
700 |
+
str(scaling_factor)
|
701 |
+
)
|
702 |
|
703 |
# ---------------------
|
704 |
# Gradio Interface
|