-
Notifications
You must be signed in to change notification settings - Fork 9.5k
/
metafile.yml
32 lines (31 loc) · 963 Bytes
/
metafile.yml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Collections:
- Name: CondInst
Metadata:
Training Data: COCO
Training Techniques:
- SGD with Momentum
- Weight Decay
Training Resources: 8x A100 GPUs
Architecture:
- FPN
- FCOS
- ResNet
Paper: https://arxiv.org/abs/2003.05664
README: configs/condinst/README.md
Models:
- Name: condinst_r50_fpn_ms-poly-90k_coco_instance
In Collection: CondInst
Config: configs/condinst/condinst_r50_fpn_ms-poly-90k_coco_instance.py
Metadata:
Training Memory (GB): 4.4
Iterations: 90000
Results:
- Task: Object Detection
Dataset: COCO
Metrics:
box AP: 39.8
- Task: Instance Segmentation
Dataset: COCO
Metrics:
mask AP: 36.0
Weights: https://download.openmmlab.com/mmdetection/v3.0/condinst/condinst_r50_fpn_ms-poly-90k_coco_instance/condinst_r50_fpn_ms-poly-90k_coco_instance_20221129_125223-4c186406.pth