Rethinking ImageNet Pre-training

Kaiming HeRoss GirshickPiotr Dollár

   Papers with code   Abstract  PDF

We report competitive results on object detection and instance segmentation on the COCO dataset using standard models trained from random initialization. The results are no worse than their ImageNet pre-training counterparts even when using the hyper-parameters of the baseline system (Mask R-CNN) that were optimized for fine-tuning pre-trained models, with the sole exception of increasing the number of training iterations so the randomly initialized models may converge... (read more)

Benchmarked Models

RANK
MODEL
REPO
CODE RESULT
PAPER RESULT
ε-REPRODUCED
BUILD
1
Mask R-CNN
(ResNet-101-FPN, GN, C…)
0.477
0.474