Merge pull request #28 from WEHI-ResearchComputing/maximal_detection

Maximal detection boxes increased during inference

Author: MihinP

partinet/DynamicDet/models/yolo.py

@@ -240,11 +240,12 @@ def forward_once(self, x, profile=False):
                 self.get_score = False
             if self.get_score:
                 return score
-
+        print(f"This micrograph has difficulty score {score[:, 0][0]} for threshold {self.dy_thres}\n")
         need_second = self.training or (not self.dynamic) or score[:, 0] < self.dy_thres
         need_first_head = self.training or (self.dynamic and score[:, 0] >= self.dy_thres)
 
         if need_second:
+            print(f"This micrograph has used both detectors\n")
             for m in self.model_b2:
                 if m.f == 'input':
                     x = input_x
@@ -267,6 +268,7 @@ def forward_once(self, x, profile=False):
                 y.append(x if m.i in self.save_b2 else None)  # save output
 
         if need_first_head:
+            print(f"This micrograph has used only one detector\n")
             for m in self.model_h:
                 if m.f != -1:  # if not from previous layer
                     x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layers

partinet/DynamicDet/utils/general.py

@@ -613,7 +613,7 @@ def non_max_suppression(prediction, conf_thres=0.25, iou_thres=0.45, classes=Non
 
     # Settings
     min_wh, max_wh = 2, 4096  # (pixels) minimum and maximum box width and height
-    max_det = 300  # maximum number of detections per image
+    max_det = 30000  # maximum number of detections per image
     max_nms = 30000  # maximum number of boxes into torchvision.ops.nms()
     time_limit = 10.0  # seconds to quit after
     redundant = True  # require redundant detections

partinet/DynamicDet/utils/plots.py

@@ -34,7 +34,7 @@ def hex2rgb(h):
     return [hex2rgb(h) for h in matplotlib.colors.TABLEAU_COLORS.values()]  # or BASE_ (8), CSS4_ (148), XKCD_ (949)
 
 
-def plot_one_box(x, img, color=None, label=None, line_thickness=3):
+def plot_one_box(x, img, color=None, label=None, line_thickness=6):
     # Plots one bounding box on image img
     tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1  # line/font thickness
     color = color or [random.randint(0, 255) for _ in range(3)]