README
PerceptEval
An open-source toolkit for evaluating image perception models.
🛠️ Installation
Install the latest version:
pip install git+https://github.com/xyfJASON/PerceptEval.git
📋 Supported Tasks
| Name | Task | Protocol | Doc |
|---|---|---|---|
depth-marigold | Affine-Invariant Depth Estimation | Marigold | link |
normal-dsine | Surface Normal Estimation | DSINE | link |
iid-hypersim | Intrinsic Image Decomposition (Albedo, Shading, Residual) | Hypersim | link |
semseg-ade20k | Semantic Segmentation | ADE20K | link |
📖 Examples
We provide several examples showing how to evaluate image perception models with our PerceptEval toolkit. Please refer to the following links for detailed instructions:
| Method | Task | Doc |
|---|---|---|
| Marigold | depth-marigold, normal-dsine, iid-hypersim | link |
| DSINE | normal-dsine | link |
| Lotus | depth-marigold, normal-dsine | link |
| ColorfulShading | iid-hypersim | link |
| OneFormer | semseg-ade20k | link |
| PixWizard | depth-marigold, normal-dsine, semseg-ade20k | link |
| OmniGen | depth-marigold, normal-dsine | link |
| OneDiffusion | depth-marigold | link |
| Jodi | depth-marigold, normal-dsine, iid-hypersim | link |
📄 License
This project is licensed under the Apache License 2.0.
Some portions of the code are adapted from third-party repositories and retain their original licenses.
🖋️ Citation
If you find this project useful, please consider to 🌟star this repo. Thanks for your support!
@misc{percepteval2026,
author = {Xu, Yifeng},
title = {PerceptEval: An open-source toolkit for evaluating image perception models},
url = {https://github.com/xyfJASON/PerceptEval},
year = {2026}
}
Contributors
12 commits
Similar repos
Languages
Python
100.0%
README
PerceptEval
An open-source toolkit for evaluating image perception models.
🛠️ Installation
Install the latest version:
pip install git+https://github.com/xyfJASON/PerceptEval.git
📋 Supported Tasks
| Name | Task | Protocol | Doc |
|---|---|---|---|
depth-marigold | Affine-Invariant Depth Estimation | Marigold | link |
normal-dsine | Surface Normal Estimation | DSINE | link |
iid-hypersim | Intrinsic Image Decomposition (Albedo, Shading, Residual) | Hypersim | link |
semseg-ade20k | Semantic Segmentation | ADE20K | link |
📖 Examples
We provide several examples showing how to evaluate image perception models with our PerceptEval toolkit. Please refer to the following links for detailed instructions:
| Method | Task | Doc |
|---|---|---|
| Marigold | depth-marigold, normal-dsine, iid-hypersim | link |
| DSINE | normal-dsine | link |
| Lotus | depth-marigold, normal-dsine | link |
| ColorfulShading | iid-hypersim | link |
| OneFormer | semseg-ade20k | link |
| PixWizard | depth-marigold, normal-dsine, semseg-ade20k | link |
| OmniGen | depth-marigold, normal-dsine | link |
| OneDiffusion | depth-marigold | link |
| Jodi | depth-marigold, normal-dsine, iid-hypersim | link |
📄 License
This project is licensed under the Apache License 2.0.
Some portions of the code are adapted from third-party repositories and retain their original licenses.
🖋️ Citation
If you find this project useful, please consider to 🌟star this repo. Thanks for your support!
@misc{percepteval2026,
author = {Xu, Yifeng},
title = {PerceptEval: An open-source toolkit for evaluating image perception models},
url = {https://github.com/xyfJASON/PerceptEval},
year = {2026}
}
Similar repos
Contributors
12 commits
Languages
Python
100.0%