Releases: chaofengc/IQA-PyTorch
pyiqa v0.1.11
📢 Fix Bugs
✨ New features
You can now use pyiqa
in terminal like this:
# list all available metrics
pyiqa -ls
# test with default settings
pyiqa [metric_name(s)] --target [image_path or dir] --ref [image_path or dir]
- ✨ add wadiqam pretrained models
- ✨ add qalign
- 🔨 add scandir_images func
- 🚩 add inception_score
- 🚩 add console entry point with pyiqa command
🛠️ Improvements
🍻 New Contributors
Many thanks to the valuable contributions 🤗 !
Full Changelog: v0.1.10...v0.1.11
pyiqa v0.1.10
📢 Fix Bugs
- fix vertical flip in data transforms 872e6ca
- fix bug of fid calculation for different sizes 2da824d
- fix maniqa hyperparameter error 33a2829
- fix device handling 6b1547f
✨ New features
- Add huggingface link to download datasets 031fafd
- Add
liqe
liqe_mix
by @zwx8981 - Add
nima-koniq
andnima-spaq
bd385c7
🛠️ Improvements
- Improve training pipeline 43fb392
- Add more error messages in fid c36d31c
- Improve error message for FR and demo codes 921bd75
🍻 New Contributors
- Improve SSIM numerical stability by transforming operations into float64 by @Luciennnnnnn in #125
- Add LIQE by @zwx8981 in #127
Many thanks to their valuable contributions 🤗 !
Full Changelog: v0.1.8...v0.1.10
pyiqa v0.1.8
📢 Fix Bugs
- Fix lower_better for FID be5d8c7
- Change
mad
optiontest_y_channel
to be false by default cf84ffb - Set read rgb to
true
in inference model to force read RGB image ad93cd4
✨ New features
- Add
topiq_fr
,topiq_nr
,topiq_nr-face
,topiq_iaa
etc. introduced by our paper TOPIQ - Add
st-lpips
by @abhijay9 in #93. Many thanks ❤️ ! - Add
laion_aes
introduced by LAION-Aesthetics_Predictor - Add datasets PIQ2023, GFIQA f46faae
🛠️ Improvements
- Add documents: https://iqa-pytorch.readthedocs.io
- Update to torchvision>=0.13, torch>=1.12
- Improve dataset api 6fbae36
- Meta information files for training can be download automatically now.
- Update evaluation protocol and results 2770a7e
- Update results of
maniqa
7ee5ea6 - Add assertion for
brisque
to force gray input 4e2c707
New Contributors
Full Changelog: v0.1.7...v0.1.8
pyiqa v0.1.7
📢 Important Changes & Bug Fix
- Fix verbose option in FID 49b2297
- Fix grad backpropagation for
as_loss=True
e027618 - Fix
niqe
with gray scale input 212ecef
✨ New features
- Add metric
uranker
b142d1c - Add metric
maniqa-koniq, maniqa-kadid
fe95923 - Add metric
clipscore
for image-caption matching ecb3e5e - Add metric
entropy
to calculate gray scale image entropy like matlab 5f6d4fb - Add pytest cases for results calibration, datasets loading and gradient backward. bc5e135
🛠️ Improvements
- Recursively find images in folder for FID calculation d7ade54
- Add metric output doc in model cards
Hotfix of NRQM & PI
🚨 Hotfix: NRQM & PI Calculation Bug Resolved
We have identified and resolved a critical bug in the NRQM calculation within our toolbox. The issue stemmed from the SSIM (Structural Similarity Index) function, where only the structure similarity score was being utilized. We apologize for any inconvenience this may have caused.
As a result of this bug, the PI, which is determined by the formula PI = 0.5 * (10 - NRQM + NIQE), was also affected.
With this hotfix, we have:
- Updated the NRQM calculation to correctly incorporate all relevant components of the SSIM function.
- Adjusted the PI calculation to reflect the corrected NRQM values.
We strongly recommend using the latest release to benefit from these crucial fixes.
If you encounter any issues or have further questions, please don't hesitate to reach out to our support team. Thank you for your understanding and continued support.
IQA-PyTorch v0.1.6
⚠️ Important Changes & Bug Fix
- Fix OOM on GPU for NRQM fcb7f6e
- Fix device problem for gradient calculation 08f8850
- Fix bug for small image test of NIQE 8d7462d
- Fix default dataset config 1baa70e
- Fix clip installation error 53d176f
- feat: add
psnry
for y colorspace;ssimc
for RGBssim
New features
- Add CNNIQA, TreS
- Pass forward argument to inference model
- Add loss reduction when using metric as loss f03d7f1
Improvements
- Update benchmark results
- Update clipiqa+
IQA-PyTorch v0.1.5
⚠️ Fix bugs
- Fix FID bug
- Fix read meta info error in livechallenge.
- Fix shape error for NRQM
- Fix bug in nancov
- Add missing requirements package
- Fix link for lpips squeeze net version
New features
- Add MANIQA, AHIQ pretrained weights
- Add
metric_mode
option forlist_models
- Add new metrics: FID, MANIQA
- Enable image path as inputs. See demo codes in README
- Add
as_loss
option to enable gradient backpropagation for metric. DefaultFalse
.
Improvements
- Use
epoch
instead ofiteration
in lr scheduler - Add
clean_state_dict
before loading pretrain model
IQA-PyTorch v0.1.4
New features
- Add new metrics: FID, MANIQA
- Enable image path as inputs. See demo codes in README
- Add
as_loss
option to enable gradient backpropagation for metric. DefaultFalse
.
Fix bugs
- Fix rmse error
- Fix benchmark test with PieAPP
Improvements
- Disable gradient calculation by default for convenience.
- Add
filter2
function to matlab utils - Add
reduction
option to EMDLoss - Add
crop_border
option to PSNR, SSIM
pyiqa v0.1.3 beta version
New features
- Add RMSE metric
- Add scale fitting option for calculation of PLCC and RMSE
Fix bugs
- Fix NIQE error when calculating images with large (>96 x 96) plain regions (regions with constant value). See #23
- Correct batch inference error for pieapp
- Fix compatibility of "torch.linalg.svd" for pytorch 1.9 #25
Improvements
- Improve function interface to match original matlab codes, including
nanmean
,nancov
,blockproc
,fspecial
. - Improve efficiency of symmetric padding, according to this link
- For pieapp, we change default stride to 27 for computation-performance trade off.
IQA-PyTorch v0.1.3 Alpha version
New features
- We add the following new metrics:
- pieapp
- paq2piq
- dbcnn trained with our own splits and configurations
- Add SRCC based loss function
Important change
We change the default musiq
weights from musiq-ava
to musiq-koniq
because it is more robust according to NR benchmark results
Fix bugs
- Remove
Lambda
transform in dataset to enable distributed training - Fix paq2piq batch test error