From eb0e14d4c572f90b956e01864f11bacbe20b5782 Mon Sep 17 00:00:00 2001 From: Taehun Kim Date: Tue, 15 Nov 2022 13:27:00 +0900 Subject: [PATCH] readme update --- README.md | 48 ++++++++++++++++++++++++------------------------ 1 file changed, 24 insertions(+), 24 deletions(-) diff --git a/README.md b/README.md index 9b8ea64..0172303 100644 --- a/README.md +++ b/README.md @@ -48,11 +48,11 @@ To appear in the 16th Asian Conference on Computer Vision (ACCV2022) 5. [Getting Started](#flight_departure-getting-started) 6. [Model Zoo](#giraffe-model-zoo) 7. [Results](#100-results) - * [Quantitative Results](#quantitative-results) - * [Qualitative Results](#qualitative-results) + * [Quantitative Results](#quantitative-results) + * [Qualitative Results](#qualitative-results) 8. [Citation](#trophy-citation) 9. [Acknowledgement](#mortar_board-acknowledgement) - * [Special Thanks to](#tada-special-thanks-to) + * [Special Thanks to](#tada-special-thanks-to) 10. [References](#paperclip-references) ## :newspaper: News @@ -117,33 +117,33 @@ Please refer to [model_zoo.md](./docs/model_zoo.md) for downloading pre-trained ### :new: [Added in 2022.10.20] Quantitative Results on DIS5K Dataset [[Log file](https://postechackr-my.sharepoint.com/:t:/g/personal/taehoon1018_postech_ac_kr/EeczZ1XEboZKhxqif9m1VwsBhMc--dLYqlZ_5TicEXr2ZA?e=aCFXhp)] -
Click to see the results -

+

Click to see the results +

- * Trained with LR scale (384 X 384) [[Config](./configs/extra_dataset/InSPyReNet_SwinB_DIS5K_LR.yaml)] + * Trained with LR scale (384 X 384) [[Config](./configs/extra_dataset/InSPyReNet_SwinB_DIS5K_LR.yaml)] - Dataset | Sm | mae | adpEm | maxEm | avgEm | adpFm | maxFm | avgFm | wFm | mBA | HCE* | - :- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- | - DIS-VD | 0.8868 | 0.0427 | 0.9145 | 0.9352 | 0.9217 | 0.8295 | 0.8760 | 0.8523 | 0.8259 | 0.7654 | 905 | - DIS-TE1 | 0.8618 | 0.0447 | 0.8679 | 0.9071 | 0.8952 | 0.7556 | 0.8341 | 0.8083 | 0.7771 | 0.7453 | 148 | - DIS-TE2 | 0.8934 | 0.0383 | 0.9131 | 0.9356 | 0.9253 | 0.8281 | 0.8811 | 0.8599 | 0.8339 | 0.7587 | 316 | - DIS-TE3 | 0.9019 | 0.0381 | 0.9278 | 0.9496 | 0.9376 | 0.8529 | 0.9038 | 0.8802 | 0.8558 | 0.7741 | 582 | - DIS-TE4 | 0.8913 | 0.0461 | 0.9316 | 0.9433 | 0.9255 | 0.8545 | 0.8915 | 0.8655 | 0.8395 | 0.7789 | 2243 | + Dataset | Sm | mae | adpEm | maxEm | avgEm | adpFm | maxFm | avgFm | wFm | mBA | HCE* | + :- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- | + DIS-VD | 0.8868 | 0.0427 | 0.9145 | 0.9352 | 0.9217 | 0.8295 | 0.8760 | 0.8523 | 0.8259 | 0.7654 | 905 | + DIS-TE1 | 0.8618 | 0.0447 | 0.8679 | 0.9071 | 0.8952 | 0.7556 | 0.8341 | 0.8083 | 0.7771 | 0.7453 | 148 | + DIS-TE2 | 0.8934 | 0.0383 | 0.9131 | 0.9356 | 0.9253 | 0.8281 | 0.8811 | 0.8599 | 0.8339 | 0.7587 | 316 | + DIS-TE3 | 0.9019 | 0.0381 | 0.9278 | 0.9496 | 0.9376 | 0.8529 | 0.9038 | 0.8802 | 0.8558 | 0.7741 | 582 | + DIS-TE4 | 0.8913 | 0.0461 | 0.9316 | 0.9433 | 0.9255 | 0.8545 | 0.8915 | 0.8655 | 0.8395 | 0.7789 | 2243 | - * Trained with HR scale (1024 X 1024) [[Config](./configs/extra_dataset/InSPyReNet_SwinB_DIS5K.yaml)] + * Trained with HR scale (1024 X 1024) [[Config](./configs/extra_dataset/InSPyReNet_SwinB_DIS5K.yaml)] - Dataset | Sm | mae | adpEm | maxEm | avgEm | adpFm | maxFm | avgFm | wFm | mBA | HCE* | - :- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- | - DIS-VD | 0.8997 | 0.0416 | 0.9141 | 0.9398 | 0.9216 | 0.8364 | 0.8893 | 0.8603 | 0.8336 | 0.8003 | 904 | - DIS-TE1 | 0.8733 | 0.0432 | 0.8743 | 0.9070 | 0.8940 | 0.7771 | 0.8453 | 0.8195 | 0.7882 | 0.7966 | 110 | - DIS-TE2 | 0.9053 | 0.0364 | 0.9154 | 0.9410 | 0.9279 | 0.8439 | 0.8941 | 0.8703 | 0.8462 | 0.8034 | 255 | - DIS-TE3 | 0.9179 | 0.0341 | 0.9403 | 0.9586 | 0.9427 | 0.8752 | 0.9191 | 0.8923 | 0.8707 | 0.8079 | 522 | - DIS-TE4 | 0.9052 | 0.0424 | 0.9361 | 0.9535 | 0.9282 | 0.8598 | 0.9051 | 0.8700 | 0.8479 | 0.7994 | 2336 | + Dataset | Sm | mae | adpEm | maxEm | avgEm | adpFm | maxFm | avgFm | wFm | mBA | HCE* | + :- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- |:- | + DIS-VD | 0.8997 | 0.0416 | 0.9141 | 0.9398 | 0.9216 | 0.8364 | 0.8893 | 0.8603 | 0.8336 | 0.8003 | 904 | + DIS-TE1 | 0.8733 | 0.0432 | 0.8743 | 0.9070 | 0.8940 | 0.7771 | 0.8453 | 0.8195 | 0.7882 | 0.7966 | 110 | + DIS-TE2 | 0.9053 | 0.0364 | 0.9154 | 0.9410 | 0.9279 | 0.8439 | 0.8941 | 0.8703 | 0.8462 | 0.8034 | 255 | + DIS-TE3 | 0.9179 | 0.0341 | 0.9403 | 0.9586 | 0.9427 | 0.8752 | 0.9191 | 0.8923 | 0.8707 | 0.8079 | 522 | + DIS-TE4 | 0.9052 | 0.0424 | 0.9361 | 0.9535 | 0.9282 | 0.8598 | 0.9051 | 0.8700 | 0.8479 | 0.7994 | 2336 | - * *: HCE here is relax Human Correction Error which is proposed in DIS([project page](https://xuebinqin.github.io/dis/index.html) | [paper](https://arxiv.org/pdf/2203.03041.pdf)) and you can compute yourself from their [github](https://github.com/xuebinqin/DIS) repository. + * *: HCE here is relax Human Correction Error which is proposed in DIS([project page](https://xuebinqin.github.io/dis/index.html) | [paper](https://arxiv.org/pdf/2203.03041.pdf)) and you can compute yourself from their [github](https://github.com/xuebinqin/DIS) repository. -

-
+

+
### Qualitative Results