Cascade R-CNN: High Quality Object Detection and Instance Segmentation

Zhaowei CaiNuno Vasconcelos

   Papers with code   Abstract  PDF

In object detection, the intersection over union (IoU) threshold is frequently used to define positives/negatives. The threshold used to train a detector defines its \textit{quality}... (read more)

Benchmarked Models

RANK
MODEL
REPO
CODE RESULT
PAPER RESULT
ε-REPRODUCED
BUILD
1
Cascade Mask R-CNN
(ResNeXt-101 64x4d-FPN…)
0.457
0.458
2
Cascade Mask R-CNN
(ResNeXt-101 64x4d-FPN…)
0.452
--
3
Cascade Mask R-CNN
(ResNeXt-101 32x4d-FPN…)
0.447
--
4
Cascade R-CNN
(ResNeXt-101 64x4d-FPN…)
0.447
--
5
Cascade R-CNN
(ResNeXt-101 64x4d-FPN…)
0.445
--
6
Cascade Mask R-CNN
(ResNeXt-101 32x4d-FPN…)
0.444
--
7
Cascade R-CNN
(ResNeXt-101 32x4d-FPN…)
0.440
--
8
Cascade R-CNN
(ResNeXt-101 32x4d-FPN…)
0.436
--
9
Cascade Mask R-CNN
(ResNet-101-FPN, 20e LR)
0.433
0.433
10
Cascade Mask R-CNN
(ResNet-101-FPN, 1x LR)
0.426
--
11
Cascade R-CNN
(ResNet-101-FPN+, casc…)
0.425
--
12
Cascade Mask R-CNN
(ResNet-50-FPN, 20e LR)
0.423
--
13
Cascade R-CNN
(ResNet-101-FPN, 1x LR)
0.420
--
14
Cascade Mask R-CNN
(ResNet-50-FPN, 1x LR)
0.412
0.413
15
Cascade R-CNN
(ResNet-50-FPN, 20e LR)
0.411
--
16
Cascade R-CNN
(ResNet-50-FPN+)
0.404
--