Papers › Fixing the train-test resolution discrepancy: FixEfficientNet

Fixing the train-test resolution discrepancy: FixEfficientNet

18 Mar 2020arXiv:2003.08237archive 2025-07-28

Hugo Touvron, Andrea Vedaldi, Matthijs Douze, Hervé Jégou

This paper provides an extensive analysis of the performance of the EfficientNet image classifiers with several recent training procedures, in particular one that corrects the discrepancy between train and test images. The resulting network, called FixEfficientNet, significantly outperforms the initial architecture with the same number of parameters. For instance, our FixEfficientNet-B0 trained without additional training data achieves 79.3% top-1 accuracy on ImageNet with 5.3M parameters. This is a +0.5% absolute improvement over the Noisy student EfficientNet-B0 trained with 300M unlabeled images. An EfficientNet-L2 pre-trained with weak supervision on 300M unlabeled images and further optimized with FixRes achieves 88.5% top-1 accuracy (top-5: 98.7%), which establishes the new state of the art for ImageNet with a single crop. These improvements are thoroughly evaluated with cleaner protocols than the one usually employed for Imagenet, and particular we show that our improvement remains in the experimental setting of ImageNet-v2, that is less prone to overfitting, and with ImageNet Real Labels. In both cases we also establish the new state of the art.

PaperPDFCode

In Syntology View this paper on Syntology, its page in Syntology's graph. That page lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

facebookresearch/FixRes officialmentioned in paperpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Fixing the train-test resolution discrepancy (arXiv:1906.06423; that paper's own run record: 0 ran · 2 unverified); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Data AugmentationImage Classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification ImageNet FixEfficientNet-L2 GFLOPs 585 #44 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-L2 Number of params 480M #44 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-L2 Top 1 Accuracy 88.5% #44 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B7 GFLOPs 82 #104 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B7 Number of params 66M #104 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B7 Top 1 Accuracy 87.1% #104 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B6 Number of params 43M #128 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B6 Top 1 Accuracy 86.7% #128 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B5 Number of params 30M #146 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B5 Top 1 Accuracy 86.4% #146 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B4 Number of params 19M #186 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B4 Top 1 Accuracy 85.9% #186 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B8 Top 1 Accuracy 85.7% #206 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B3 Number of params 12M #266 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B3 Top 1 Accuracy 85% #266 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNetB4 Number of params 19M #363 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNetB4 Top 1 Accuracy 84.0% #363 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B2 Number of params 9.2M #411 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B2 Top 1 Accuracy 83.6% #411 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B1 Number of params 7.8M #522 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B1 Top 1 Accuracy 82.6% #522 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B0 GFLOPs 1.60 #713 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B0 Number of params 5.3M #713 of 1060 Archive leaderboard report
Image Classification ImageNet FixEfficientNet-B0 Top 1 Accuracy 80.2% #713 of 1060 Archive leaderboard report
Image Classification ImageNet ReaL FixEfficientNet-L2 Accuracy 90.9% #10 of 57 Archive leaderboard report
Image Classification ImageNet ReaL FixEfficientNet-L2 Params 480M #10 of 57 Archive leaderboard report
Image Classification ImageNet ReaL FixEfficientNet-B8 Accuracy 90.0% #21 of 57 Archive leaderboard report
Image Classification ImageNet ReaL FixEfficientNet-B8 Params 87M #21 of 57 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

1x1 ConvolutionAverage PoolingBatch NormalizationColorJitterConvolutionDense ConnectionsDepthwise ConvolutionDepthwise Separable ConvolutionDropoutEfficientNetFixResInverted Residual BlockLabel SmoothingPointwise ConvolutionRMSPropRandom Horizontal FlipRandom Resized CropReLUSigmoid ActivationSqueeze-and-Excitation Block

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections