search

facebookresearch / moco

PyTorch implementation of MoCo: https://arxiv.org/abs/1911.05722
MIT License
4.83k stars 794 forks source link

About the FPN setting on COCO #39

Closed whai362 closed 4 years ago

whai362 commented 4 years ago

Hi, thanks for open-sourcing this excellent repo. I try to reproduce the performance of Mask R-CNN (R50-FPN, 1x) following https://github.com/facebookresearch/moco/issues/34#issuecomment-633177648. But there still a gap between the reproduced AP (34.8%) and the AP reported in the paper (35.1%). Are there differences between our config file and yours? This is our config file:

_BASE_: "Base-RCNN-FPN.yaml"
MODEL:
  PIXEL_MEAN: [123.675, 116.280, 103.530]
  PIXEL_STD: [58.395, 57.120, 57.375]
  MASK_ON: True
  WEIGHTS: "Mocov1 Model"
  BACKBONE:
    FREEZE_AT: 0
  RESNETS:
    DEPTH: 50
    NORM: "SyncBN"
    STRIDE_IN_1X1: False
  FPN:
    NORM: "SyncBN"
TEST:
  PRECISE_BN:
    ENABLED: True
  EVAL_PERIOD: 5000
SOLVER:
  STEPS: (60000, 80000)
  MAX_ITER: 90000
INPUT:
  FORMAT: "RGB"
OUTPUT_DIR: "./output/mask_fpn_1x_mocov1/"
KaimingHe commented 4 years ago

A gap of 0.3 AP is within a reasonable range of random variation in COCO.