text
stringlengths
0
820
codebases and pretraining on FMoW dataset for 800 epochs. The
results differ from their reported results, but are evaluated consis-
tently with ours. * Reports the results from the SatMAE paper [13].
Semantic segmentation transfer We use the SpaceNet v1
building segmentation dataset [53] to evaluate semantic seg-
mentation results on contrastive and MAE-based pretrainingmethods. Prior methods relied on the PSANet [68] segmen-
tation architecture, while Scale-MAE uses the UperNet [58]
segmentation architecture which is more common for ViT
backbones. For even comparison, we test the current state-
of-the-art SatMAE and ConvMAE methods with UperNet
as well. Results are detailed in Table 4.
Method Backbone Model mIoU
Sup. (Scratch) ResNet50 PSANet 75.6
GASSL [3] ResNet50 PSANet 78.5
Sup. (Scratch) ViT-Large PSANet 74.7
SatMAE [13] ViT-Large PSANet 78.1
Sup. (Scratch) ViT-Large UperNet 71.6
Vanilla MAE ViT-Large UperNet 77.9
SatMAE ViT-Large UperNet 78.0
ConvMAE ViT-Large UperNet 77.6
Scale-MAE ViT-Large UperNet 78.9
Table 4. Semantic segmentation results on SpaceNet v1.
Scale-MAE outperforms other methods across backbone and seg-
mentation architectures, where Sup. (Scratch) indicates a super-
vised model trained from scratch (a randomly initialized network).
With the same pretraining settings, Scale-MAE outper-
forms SatMAE by 0.9 mIoU, ConvMAE by 1.3 mIoU, and
a vanilla MAE by 1.0 mIoU. Scale-MAE outperforms all
other prior work, including GASSL [3], which SatMAE did
not outperform on the mean Intersection over Union (mIoU)
metric for semantic segmentation. Particularly, Scale-MAE
increases the gap in performance as the resolution of input
imagery becomes coarser, highlighting the absolute scale-
invariance introduced by our method.
In Figure 6, we compare SpaceNet v1 evaluations across
downscaled images (at 50%, 75%, and 100% of the origi-
nal image size) for Scale-MAE , SatMAE, and ConvMAE.
Similar to the classification results, Scale-MAE maintains
higher semantic segmentation performance over both meth-
ods, even with images at a coarser GSD. In fact, the per-
formance gap grows at coarser GSDs. Compared to the
next-best-performing method at the input GSD, Scale-MAE
is 0.9 mIoU higher, at 75% GSD Scale-MAE is 1.2 mIoU
higher, and at 50% Scale-MAE is 1.7 mIoU higher.
In Table 5, we further evaluate Scale-MAE , SatMAE, and
ConvMAE across SpaceNet v1, SpaceNet v2 [53], INRIA
Aerial Image [44], and GID-15 [59] remote sensing datasets
at native resolution. Scale-MAE outperforms both compara-
ble methods across all benchmarks.
4.2. Ablations
We ablate the key components of the Scale-MAE pretrain-
ing framework. For these experiments, we use a lightweight
pretraining setting, where we pretrain for 300 epochs on
50% 75% 100%
Relative GSD7072747678mIoU
Scale-MAE
SatMAE
ConvMAEFigure 6. SpaceNet v1 evaluation across downscaled images for
both Scale-MAE and SatMAE. Scale-MAE maintains higher se-
mantic segmentation performance over SatMAE, even with images
of coarser GSD.
SN1 SN2 INR. G15
RI SH VE PA KH - -
Conv. 77.6 78.7 82.2 78.3 74.8 82.2 37.4
Sat. 78.0 81.9 86.6 80.3 76.1 83.0 44.3
Scale 78.9 82.2 87.4 81.1 77.1 84.2 46.2
Table 5. mIoU on semantic segmentation tasks. SN1/2 (SpaceNet
v1/2), RI: Rio, SH: Shanghai, VE: Vegas, PA: Paris, KH: Khar-
toum; INR: INRIA; G15: GID-15. Conv., Sat., and Scale. are
ConvMAE, SatMAE, and Scale-MAE.
Method GSDPE KNN 50% KNN 100%
Vanilla MAE 72.8 77.8
Vanilla MAE ! 75.4 78.5
MAE + LP 75.3 79.6
Scale-MAE ! 78.1 80.7
Table 6. Ablation results indicating the importance of GSDPE as
determined by a KNN classification on RESISC-45 at a relative
GSD of 50% and 100% of its native GSD. Using the GSDPE
leads to better performance for both Scale-MAE and the Vanilla
MAE. MAE + LP denotes the vanilla MAE with the addition of
our progressive Laplacian decoder.
FMoW (rather than 800) and use a ViT-Base encoder (rather
than ViT-Large), and evaluate using a kNN evaluation on
RESISC-45 at 100% and 50% of its native GSD. The key
contributions that we ablate are as follows: the GSD posi-
tional encoder in Table 6, in which we find that the GSD
postional encoder benefits both Scale-MAE and Vanilla MAE
across resolutions. In Table 8, we see that the number of
transformer layers can be reduced from 8 to 3 compared to a
Vanilla MAE, which results in a performance improvement.
The standard masking rate of 75% still appears optimal for
Scale-MAE according to the results in Table 7.
In Table 9 we ablate the necessity of the low and high res-
olution reconstructions. Specifically, we test reconstructing
the low resolution image only, the high resolution image, andMask Rate KNN 50% KNN 100%
70% 77.3 79.3
75% 78.1 80.7
80% 78.1 79.9
Table 7. Ablation results indicating that a 75% mask rate is optimal
as determined by a KNN classification on RESISC-45 at a relative