# Out-of-Distribution Detection on CIFAR-10 (FPR/AUROC vs SVHN, LSUN, iSUN, Texture, Places365)

Source: [Out-of-Distribution Detection on CIFAR-10 (FPR/AUROC vs SVHN, LSUN, iSUN, Texture, Places365)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:44:54.062Z

Evaluation of Out-of-Distribution Detection performance on CIFAR-10, reporting both False Positive Rate (FPR) and Area Under the ROC Curve (AUROC) against a suite of OOD datasets including SVHN, LSUN, iSUN, Texture, and Places365, alongside In-Distribution Accuracy (ID ACC).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Out-of-Distribution Detection |
| Dataset (including any stated split) | CIFAR-10 |
| Metric | FPR (SVHN) |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 19 of 19 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [FPR (SVHN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR%20(SVHN)>): Lower is better; unit: PERCENT.

- [AUROC (SVHN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(SVHN)>): Higher is better; unit: PERCENT.

- [FPR (LSUN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR%20(LSUN)>): Lower is better; unit: PERCENT.

- [AUROC (LSUN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(LSUN)>): Higher is better; unit: PERCENT.

- [FPR (iSUN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR%20(iSUN)>): Lower is better; unit: PERCENT.

- [AUROC (iSUN)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(iSUN)>): Higher is better; unit: PERCENT.

- [FPR (Texture)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR%20(Texture)>): Lower is better; unit: PERCENT.

- [AUROC (Texture)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(Texture)>): Higher is better; unit: PERCENT.

- [FPR (Places365)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR%20(Places365)>): Lower is better; unit: PERCENT.

- [AUROC (Places365)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(Places365)>): Higher is better; unit: PERCENT.

- [Average FPR](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=Average%20FPR>): Lower is better; unit: PERCENT.

- [Average AUROC](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=Average%20AUROC>): Higher is better; unit: PERCENT.

- [ID Accuracy](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=ID%20Accuracy>): Higher is better; unit: PERCENT.

- [FPR95 (LSUN-C)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR95%20(LSUN-C)>): Lower is better; unit: PERCENT.

- [AUROC (LSUN-C)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(LSUN-C)>): Higher is better; unit: PERCENT.

- [FPR95 (LSUN-R)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=FPR95%20(LSUN-R)>): Lower is better; unit: PERCENT.

- [AUROC (LSUN-R)](<https://www.sota2.com/research/sota/out-of-distribution-detection-on-cifar-10-fpr-auroc-vs-svhn-lsun-isun-texture-places365.md?metric=AUROC%20(LSUN-R)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | FPR (SVHN) | Result date | Report paper |
| --- | --- | --- | --- |
| SSD+ | 1.51 | Not recorded | 68fa6e2f1e209131cea3e69b |
| KNN+ | 2.42 | Not recorded | 68fa6e2f1e209131cea3e69b |
| Mahalanobis | 6.42 | Not recorded | 68fa6f721e209131cea45a18 |
| Mahalanobis | 9.24 | Not recorded | 68fa6e2f1e209131cea3e69b |
| KNN | 13.51 | Not recorded | 68fa6f721e209131cea45a18 |
| VRA+ | 13.54 | Not recorded | 68fa6f721e209131cea45a18 |
| GODIN | 18.72 | Not recorded | 68fa6e2f1e209131cea3e69b |
| VRA | 18.75 | Not recorded | 68fa6f721e209131cea45a18 |
| ODIN | 25.29 | Not recorded | 68fa6f721e209131cea45a18 |
| DICE | 25.99 | Not recorded | 68fa6f721e209131cea45a18 |
| KNN | 27.97 | Not recorded | 68fa6e2f1e209131cea3e69b |
| SHE | 28.12 | Not recorded | 68fa6f721e209131cea45a18 |
| CSI | 37.38 | Not recorded | 68fa6e2f1e209131cea3e69b |
| Energy | 40.61 | Not recorded | 68fa6f721e209131cea45a18 |
| ReAct | 41.64 | Not recorded | 68fa6f721e209131cea45a18 |
| MSP | 47.27 | Not recorded | 68fa6f721e209131cea45a18 |
| ODIN | 53.78 | Not recorded | 68fa6e2f1e209131cea3e69b |
| Energy | 54.41 | Not recorded | 68fa6e2f1e209131cea3e69b |
| MSP | 59.66 | Not recorded | 68fa6e2f1e209131cea3e69b |

### SSD+

FPR (SVHN): 1.51. Result ID: 6990142f3a6c314849fdfb04.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa698f1e209131cea22a16.md>)

[Implementation](<https://github.com/inspire-group/SSD>)

Recorded evaluation configuration: {"Contrastive Learning":"Yes","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### KNN+

FPR (SVHN): 2.42. Result ID: 6990142f3a6c314849fdfb05.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Implementation](<https://github.com/deeplearning-wisc/knn-ood>)

Recorded evaluation configuration: {"Contrastive Learning":"Yes","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### Mahalanobis

FPR (SVHN): 6.42. Result ID: 6990d1d43a6c3148490388f6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64f91e209131cea077ac.md>)

[Implementation](<https://github.com/pokaxpoka/deep_Mahalanobis_detector>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### Mahalanobis

FPR (SVHN): 9.24. Result ID: 6990142f3a6c314849fdfb01.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64f91e209131cea077ac.md>)

[Implementation](<https://github.com/pokaxpoka/deep_Mahalanobis_detector>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### KNN

FPR (SVHN): 13.51. Result ID: 6990d1d43a6c3148490388f9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Implementation](<https://github.com/deeplearning-wisc/knn-ood>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### VRA+

FPR (SVHN): 13.54. Result ID: 6990d1d43a6c3148490388fd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### GODIN

FPR (SVHN): 18.72. Result ID: 6990142f3a6c314849fdfb00.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa684a1e209131cea1b26f.md>)

[Implementation](<https://github.com/facebookresearch/odin>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### VRA

FPR (SVHN): 18.75. Result ID: 6990d1d43a6c3148490388fc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### ODIN

FPR (SVHN): 25.29. Result ID: 6990d1d43a6c3148490388f5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa684a1e209131cea1b26f.md>)

[Implementation](<https://github.com/facebookresearch/odin>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### DICE

FPR (SVHN): 25.99. Result ID: 6990d1d43a6c3148490388fa.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cfc1e209131cea371c4.md>)

[Implementation](<https://github.com/deeplearning-wisc/dice.git>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### KNN

FPR (SVHN): 27.97. Result ID: 6990142f3a6c314849fdfb02.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Implementation](<https://github.com/deeplearning-wisc/knn-ood>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### SHE

FPR (SVHN): 28.12. Result ID: 6990d1d43a6c3148490388fb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### CSI

FPR (SVHN): 37.38. Result ID: 6990142f3a6c314849fdfb03.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68881e209131cea1cb66.md>)

[Implementation](<https://github.com/alinlab/CSI>)

Recorded evaluation configuration: {"Contrastive Learning":"Yes","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### Energy

FPR (SVHN): 40.61. Result ID: 6990d1d43a6c3148490388f7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69cc1e209131cea24242.md>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### ReAct

FPR (SVHN): 41.64. Result ID: 6990d1d43a6c3148490388f8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b4f1e209131cea2d8b5.md>)

[Implementation](<https://github.com/deeplearning-wisc/react.git>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### MSP

FPR (SVHN): 47.27. Result ID: 6990d1d43a6c3148490388f4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f721e209131cea45a18.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64e41e209131cea0731a.md>)

[Implementation](<https://www.sota2.com/github.com/hendrycks/error-detection>)

Recorded evaluation configuration: {"Backbone":"DenseNet-101","Pre-trained on ID data":true}. This record does not assert independent reproduction.

### ODIN

FPR (SVHN): 53.78. Result ID: 6990142f3a6c314849fdfafe.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa684a1e209131cea1b26f.md>)

[Implementation](<https://github.com/facebookresearch/odin>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### Energy

FPR (SVHN): 54.41. Result ID: 6990142f3a6c314849fdfaff.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69cc1e209131cea24242.md>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

### MSP

FPR (SVHN): 59.66. Result ID: 6990142f3a6c314849fdfafd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e2f1e209131cea3e69b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64e41e209131cea0731a.md>)

[Implementation](<https://www.sota2.com/github.com/hendrycks/error-detection>)

Recorded evaluation configuration: {"Contrastive Learning":"No","Backbone":"ResNet-18","Training Data":"CIFAR-10"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.