Skip to content

Add optional segmentation metrics plots to MetricsPlotSink. #410

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: develop
Choose a base branch
from
+111 −55
Open

Add optional segmentation metrics plots to MetricsPlotSink. #410

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
166 changes: 111 additions & 55 deletions rfdetr/util/metrics.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,65 +48,121 @@ def get_array(key):
epochs = get_array('epoch')
train_loss = get_array('train_loss')
test_loss = get_array('test_loss')
test_coco_eval = [h['test_coco_eval_bbox'] for h in self.history if 'test_coco_eval_bbox' in h]
ap50_90 = np.array([safe_index(x, 0) for x in test_coco_eval if x is not None], dtype=np.float32)
ap50 = np.array([safe_index(x, 1) for x in test_coco_eval if x is not None], dtype=np.float32)
ar50_90 = np.array([safe_index(x, 8) for x in test_coco_eval if x is not None], dtype=np.float32)

ema_coco_eval = [h['ema_test_coco_eval_bbox'] for h in self.history if 'ema_test_coco_eval_bbox' in h]
ema_ap50_90 = np.array([safe_index(x, 0) for x in ema_coco_eval if x is not None], dtype=np.float32)
ema_ap50 = np.array([safe_index(x, 1) for x in ema_coco_eval if x is not None], dtype=np.float32)
ema_ar50_90 = np.array([safe_index(x, 8) for x in ema_coco_eval if x is not None], dtype=np.float32)

fig, axes = plt.subplots(2, 2, figsize=(18, 12))
test_coco_eval_bbox = [h['test_coco_eval_bbox'] for h in self.history if 'test_coco_eval_bbox' in h]
ap50_90_bbox = np.array([safe_index(x, 0) for x in test_coco_eval_bbox if x is not None], dtype=np.float32)
ap50_bbox = np.array([safe_index(x, 1) for x in test_coco_eval_bbox if x is not None], dtype=np.float32)
ar50_90_bbox = np.array([safe_index(x, 8) for x in test_coco_eval_bbox if x is not None], dtype=np.float32)

ema_coco_eval_bbox = [h['ema_test_coco_eval_bbox'] for h in self.history if 'ema_test_coco_eval_bbox' in h]
ema_ap50_90_bbox = np.array([safe_index(x, 0) for x in ema_coco_eval_bbox if x is not None], dtype=np.float32)
ema_ap50_bbox = np.array([safe_index(x, 1) for x in ema_coco_eval_bbox if x is not None], dtype=np.float32)
ema_ar50_90_bbox = np.array([safe_index(x, 8) for x in ema_coco_eval_bbox if x is not None], dtype=np.float32)

test_coco_eval_masks = [h['test_coco_eval_masks'] for h in self.history if 'test_coco_eval_masks' in h]
ap50_90_masks = np.array([safe_index(x, 0) for x in test_coco_eval_masks if x is not None], dtype=np.float32)
ap50_masks = np.array([safe_index(x, 1) for x in test_coco_eval_masks if x is not None], dtype=np.float32)
ar50_90_masks = np.array([safe_index(x, 8) for x in test_coco_eval_masks if x is not None], dtype=np.float32)

ema_coco_eval_masks = [h['ema_test_coco_eval_masks'] for h in self.history if 'ema_test_coco_eval_masks' in h]
ema_ap50_90_masks = np.array([safe_index(x, 0) for x in ema_coco_eval_masks if x is not None], dtype=np.float32)
ema_ap50_masks = np.array([safe_index(x, 1) for x in ema_coco_eval_masks if x is not None], dtype=np.float32)
ema_ar50_90_masks = np.array([safe_index(x, 8) for x in ema_coco_eval_masks if x is not None], dtype=np.float32)

if len(test_coco_eval_masks) or len(ema_coco_eval_masks):
fig, axes = plt.subplots(4, 2, figsize=(18, 12))
else:
fig, axes = plt.subplots(2, 2, figsize=(18, 12))

# Subplot (0,0): Training and Validation Loss
# Training and Validation Loss
r, c = 0, 0
if len(epochs) > 0:
if len(train_loss):
axes[0][0].plot(epochs, train_loss, label='Training Loss', marker='o', linestyle='-')
axes[r][c].plot(epochs, train_loss, label='Training Loss', marker='o', linestyle='-')
if len(test_loss):
axes[0][0].plot(epochs, test_loss, label='Validation Loss', marker='o', linestyle='--')
axes[0][0].set_title('Training and Validation Loss')
axes[0][0].set_xlabel('Epoch Number')
axes[0][0].set_ylabel('Loss Value')
axes[0][0].legend()
axes[0][0].grid(True)

# Subplot (0,1): Average Precision @0.50
if ap50.size > 0 or ema_ap50.size > 0:
if ap50.size > 0:
axes[0][1].plot(epochs[:len(ap50)], ap50, marker='o', linestyle='-', label='Base Model')
if ema_ap50.size > 0:
axes[0][1].plot(epochs[:len(ema_ap50)], ema_ap50, marker='o', linestyle='--', label='EMA Model')
axes[0][1].set_title('Average Precision @0.50')
axes[0][1].set_xlabel('Epoch Number')
axes[0][1].set_ylabel('AP50')
axes[0][1].legend()
axes[0][1].grid(True)

# Subplot (1,0): Average Precision @0.50:0.95
if ap50_90.size > 0 or ema_ap50_90.size > 0:
if ap50_90.size > 0:
axes[1][0].plot(epochs[:len(ap50_90)], ap50_90, marker='o', linestyle='-', label='Base Model')
if ema_ap50_90.size > 0:
axes[1][0].plot(epochs[:len(ema_ap50_90)], ema_ap50_90, marker='o', linestyle='--', label='EMA Model')
axes[1][0].set_title('Average Precision @0.50:0.95')
axes[1][0].set_xlabel('Epoch Number')
axes[1][0].set_ylabel('AP')
axes[1][0].legend()
axes[1][0].grid(True)

# Subplot (1,1): Average Recall @0.50:0.95
if ar50_90.size > 0 or ema_ar50_90.size > 0:
if ar50_90.size > 0:
axes[1][1].plot(epochs[:len(ar50_90)], ar50_90, marker='o', linestyle='-', label='Base Model')
if ema_ar50_90.size > 0:
axes[1][1].plot(epochs[:len(ema_ar50_90)], ema_ar50_90, marker='o', linestyle='--', label='EMA Model')
axes[1][1].set_title('Average Recall @0.50:0.95')
axes[1][1].set_xlabel('Epoch Number')
axes[1][1].set_ylabel('AR')
axes[1][1].legend()
axes[1][1].grid(True)
axes[r][c].plot(epochs, test_loss, label='Validation Loss', marker='o', linestyle='--')
axes[r][c].set_title('Training and Validation Loss')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('Loss Value')
axes[r][c].legend()
axes[r][c].grid(True)

# BBox Average Precision @0.50
r, c = 0, 1
if ap50_bbox.size > 0 or ema_ap50_bbox.size > 0:
if ap50_bbox.size > 0:
axes[r][c].plot(epochs[:len(ap50_bbox)], ap50_bbox, marker='o', linestyle='-', label='Base Model')
if ema_ap50_bbox.size > 0:
axes[r][c].plot(epochs[:len(ema_ap50_bbox)], ema_ap50_bbox, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Precision @0.50 (BBox)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AP50')
axes[r][c].legend()
axes[r][c].grid(True)

# BBox Average Precision @0.50:0.95
r, c = 1, 0
if ap50_90_bbox.size > 0 or ema_ap50_90_bbox.size > 0:
if ap50_90_bbox.size > 0:
axes[r][c].plot(epochs[:len(ap50_90_bbox)], ap50_90_bbox, marker='o', linestyle='-', label='Base Model')
if ema_ap50_90_bbox.size > 0:
axes[r][c].plot(epochs[:len(ema_ap50_90_bbox)], ema_ap50_90_bbox, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Precision @0.50:0.95 (BBox)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AP')
axes[r][c].legend()
axes[r][c].grid(True)

# BBox Average Recall @0.50:0.95
r, c = 1, 1
if ar50_90_bbox.size > 0 or ema_ar50_90_bbox.size > 0:
if ar50_90_bbox.size > 0:
axes[r][c].plot(epochs[:len(ar50_90_bbox)], ar50_90_bbox, marker='o', linestyle='-', label='Base Model')
if ema_ar50_90_bbox.size > 0:
axes[r][c].plot(epochs[:len(ema_ar50_90_bbox)], ema_ar50_90_bbox, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Recall @0.50:0.95 (BBox)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AR')
axes[r][c].legend()
axes[r][c].grid(True)

# Masks Average Precision @0.50
r, c = 2, 0
if ap50_masks.size > 0 or ema_ap50_masks.size > 0:
if ap50_masks.size > 0:
axes[r][c].plot(epochs[:len(ap50_masks)], ap50_masks, marker='o', linestyle='-', label='Base Model')
if ema_ap50_masks.size > 0:
axes[r][c].plot(epochs[:len(ema_ap50_masks)], ema_ap50_masks, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Precision @0.50 (Masks)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AP50')
axes[r][c].legend()
axes[r][c].grid(True)

# Masks Average Precision @0.50:0.95
r, c = 2, 1
if ap50_90_masks.size > 0 or ema_ap50_90_masks.size > 0:
if ap50_90_masks.size > 0:
axes[r][c].plot(epochs[:len(ap50_90_masks)], ap50_90_masks, marker='o', linestyle='-', label='Base Model')
if ema_ap50_90_masks.size > 0:
axes[r][c].plot(epochs[:len(ema_ap50_90_masks)], ema_ap50_90_masks, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Precision @0.50:0.95 (Masks)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AP')
axes[r][c].legend()
axes[r][c].grid(True)

# Masks Average Recall @0.50:0.95
r, c = 3, 0
if ar50_90_masks.size > 0 or ema_ar50_90_masks.size > 0:
if ar50_90_masks.size > 0:
axes[r][c].plot(epochs[:len(ar50_90_masks)], ar50_90_masks, marker='o', linestyle='-', label='Base Model')
if ema_ar50_90_masks.size > 0:
axes[r][c].plot(epochs[:len(ema_ar50_90_masks)], ema_ar50_90_masks, marker='o', linestyle='--', label='EMA Model')
axes[r][c].set_title('Average Recall @0.50:0.95 (Masks)')
axes[r][c].set_xlabel('Epoch Number')
axes[r][c].set_ylabel('AR')
axes[r][c].legend()
axes[r][c].grid(True)

plt.tight_layout()
plt.savefig(f"{self.output_dir}/{PLOT_FILE_NAME}")
Expand Down