Revisiting Unreasonable Effectiveness of Data in Deep Learning Era

ICCV 2017 Chen SunAbhinav ShrivastavaSaurabh SinghAbhinav Gupta

The success of deep learning in vision can be attributed to: (a) models with high capacity; (b) increased computational power; and (c) availability of large-scale labeled data. Since 2012, there have been significant advances in representation capabilities of the models and computational capabilities of GPUs... (read more)

PDF Abstract

Evaluation results from the paper


Task Dataset Model Metric name Metric value Global rank Uses extra
training data
Compare
Image Classification ImageNet JFT-300M Finetuning Top 1 Accuracy 79.2% # 36
Image Classification ImageNet JFT-300M Finetuning Top 5 Accuracy 94.7% # 31
Semantic Segmentation PASCAL VOC 2012 ImageNet+JFT-300M Initialization Mean IoU 76.5% # 15