Spaces:
Runtime error
Runtime error
jhj0517
commited on
Commit
·
41938cd
1
Parent(s):
3c09bbc
Add point prompt
Browse files- modules/sam_inference.py +19 -6
modules/sam_inference.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
from sam2.automatic_mask_generator import SAM2AutomaticMaskGenerator
|
2 |
from sam2.build_sam import build_sam2
|
3 |
from sam2.sam2_image_predictor import SAM2ImagePredictor
|
4 |
-
from typing import Dict, List
|
5 |
import torch
|
6 |
import os
|
7 |
from datetime import datetime
|
@@ -83,7 +83,9 @@ class SamInference:
|
|
83 |
def predict_image(self,
|
84 |
image: np.ndarray,
|
85 |
model_type: str,
|
86 |
-
box: np.ndarray,
|
|
|
|
|
87 |
**params):
|
88 |
if self.model is None or self.model_type != model_type:
|
89 |
self.model_type = model_type
|
@@ -94,6 +96,8 @@ class SamInference:
|
|
94 |
try:
|
95 |
masks, scores, logits = self.image_predictor.predict(
|
96 |
box=box,
|
|
|
|
|
97 |
multimask_output=params["multimask_output"],
|
98 |
)
|
99 |
except Exception as e:
|
@@ -136,15 +140,24 @@ class SamInference:
|
|
136 |
elif input_mode == BOX_PROMPT_MODE:
|
137 |
image = image_prompt_input_data["image"]
|
138 |
image = np.array(image.convert("RGB"))
|
139 |
-
|
140 |
-
if len(
|
141 |
return [image], []
|
142 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
143 |
|
144 |
predicted_masks, scores, logits = self.predict_image(
|
145 |
image=image,
|
146 |
model_type=model_type,
|
147 |
-
box=
|
|
|
|
|
148 |
multimask_output=hparams["multimask_output"]
|
149 |
)
|
150 |
generated_masks = self.format_to_auto_result(predicted_masks)
|
|
|
1 |
from sam2.automatic_mask_generator import SAM2AutomaticMaskGenerator
|
2 |
from sam2.build_sam import build_sam2
|
3 |
from sam2.sam2_image_predictor import SAM2ImagePredictor
|
4 |
+
from typing import Dict, List, Optional
|
5 |
import torch
|
6 |
import os
|
7 |
from datetime import datetime
|
|
|
83 |
def predict_image(self,
|
84 |
image: np.ndarray,
|
85 |
model_type: str,
|
86 |
+
box: Optional[np.ndarray] = None,
|
87 |
+
point_coords: Optional[np.ndarray] = None,
|
88 |
+
point_labels: Optional[np.ndarray] = None,
|
89 |
**params):
|
90 |
if self.model is None or self.model_type != model_type:
|
91 |
self.model_type = model_type
|
|
|
96 |
try:
|
97 |
masks, scores, logits = self.image_predictor.predict(
|
98 |
box=box,
|
99 |
+
point_coords=point_coords,
|
100 |
+
point_labels=point_labels,
|
101 |
multimask_output=params["multimask_output"],
|
102 |
)
|
103 |
except Exception as e:
|
|
|
140 |
elif input_mode == BOX_PROMPT_MODE:
|
141 |
image = image_prompt_input_data["image"]
|
142 |
image = np.array(image.convert("RGB"))
|
143 |
+
prompt = image_prompt_input_data["points"]
|
144 |
+
if len(prompt) == 0:
|
145 |
return [image], []
|
146 |
+
|
147 |
+
is_prompt_point = prompt[0][-1] == 4.0
|
148 |
+
|
149 |
+
if is_prompt_point:
|
150 |
+
point_labels = np.array([1 if is_left_click else 0 for x1, y1, is_left_click, x2, y2, _ in prompt])
|
151 |
+
prompt = np.array([[x1, y1] for x1, y1, is_left_click, x2, y2, _ in prompt])
|
152 |
+
else:
|
153 |
+
prompt = np.array([[x1, y1, x2, y2] for x1, y1, is_left_click, x2, y2, _ in prompt])
|
154 |
|
155 |
predicted_masks, scores, logits = self.predict_image(
|
156 |
image=image,
|
157 |
model_type=model_type,
|
158 |
+
box=prompt if not is_prompt_point else None,
|
159 |
+
point_coords=prompt if is_prompt_point else None,
|
160 |
+
point_labels=point_labels if is_prompt_point else None,
|
161 |
multimask_output=hparams["multimask_output"]
|
162 |
)
|
163 |
generated_masks = self.format_to_auto_result(predicted_masks)
|