Skip to content

Write per class mean average precision and mean average recall into tensorboard #142

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
harmluSICKAG wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Write per class mean average precision and mean average recall into tensorboard #142

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e4d4239
feat: add per class mAP and mAR to tensorboard
harmluSICKAG Jan 13, 2025
aaafe8a
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Jan 13, 2025
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
34 changes: 34 additions & 0 deletions src/solver/det_solver.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,24 @@
from .det_engine import train_one_epoch, evaluate


def prepare_eval_metric(metric, catId, type):
# type precision: (iou, recall, cls, area range, max dets)
# type recall: (iou, cls, area range, max dets)
if type == "precision":
metric = [metric[i][j][catId][0][-1] for i in range(len(metric)) for j in range(len(metric[i]))]
elif type == "recall":
metric = [metric[i][catId][0][-1] for i in range(len(metric))]

# Filter out values <= -1
filtered_metric = [value for value in metric if value > -1]

# Calculate mean or return NaN if empty
if filtered_metric:
return sum(filtered_metric) / len(filtered_metric)
else:
return float("nan")


class DetSolver(BaseSolver):

def fit(self, ):
Expand Down Expand Up @@ -98,12 +116,28 @@ def fit(self, ):
self.device
)

coco_eval = coco_evaluator.coco_eval["bbox"]
precisions = coco_evaluator.coco_eval["bbox"].eval['precision']
recalls = coco_evaluator.coco_eval["bbox"].eval['recall']

class_results = {}
for category_id in coco_eval.cocoGt.getCatIds():
category_info = coco_eval.cocoGt.loadCats([category_id])[0]
category_name = category_info['name']
ap = prepare_eval_metric(precisions, category_id, "precision")
ar = prepare_eval_metric(recalls, category_id, "recall")
class_results[category_id] = {'name': category_name, 'mAP': ap, 'mAR': ar}

# TODO
for k in test_stats:
if self.writer and dist_utils.is_main_process():
for i, v in enumerate(test_stats[k]):
self.writer.add_scalar(f'Test/{k}_{i}'.format(k), v, epoch)

for class_id in class_results.keys():
self.writer.add_scalar(f'Test/class_{class_results[class_id]["name"]}_mAP', class_results[class_id]["mAP"], epoch)
self.writer.add_scalar(f'Test/class_{class_results[class_id]["name"]}_mAR', class_results[class_id]["mAR"], epoch)

if k in best_stat:
best_stat['epoch'] = epoch if test_stats[k][0] > best_stat[k] else best_stat['epoch']
best_stat[k] = max(best_stat[k], test_stats[k][0])
Expand Down