Skip to content

Feature/add 4k model and ci#175

Merged
li-zhi merged 54 commits intomasterfrom
feature/add-4k-model-and-ci
Jun 20, 2018
Merged

Feature/add 4k model and ci#175
li-zhi merged 54 commits intomasterfrom
feature/add-4k-model-and-ci

Conversation

@li-zhi
Copy link
Collaborator

@li-zhi li-zhi commented Jun 20, 2018

No description provided.

li-zhi added 30 commits June 10, 2018 10:40
…dd default model to BootstrapVmafQualityRunner; fix BaggingVmafQualityRunner TYPE issue.
…ents from VmafRunner::_normalize_predict_denormalize().
li-zhi added 24 commits June 17, 2018 17:36
@li-zhi li-zhi merged commit 1b0f6ef into master Jun 20, 2018
@li-zhi li-zhi deleted the feature/add-4k-model-and-ci branch June 20, 2018 07:12
slhck pushed a commit to slhck/vmaf that referenced this pull request Oct 4, 2018
* Add candidate model vmaf_4k_v0.6.1.

* Add candidate model vmaf_4k_v0.6.1.

* Add test for 4k v0.6.1 model.

* Rename model vmaf_v0.7.0rc.pkl to vmaf_4k_v0.6.1rc.pkl.

* Add bootstrap training parameter file vmaf_v6_residue_bootstrap.py; add default model to BootstrapVmafQualityRunner; fix BaggingVmafQualityRunner TYPE issue.

* Add bootstrap 4k model; add test; refactor run_testing.py.

* Change version convention of BootstrapVmafQualityRunner.

* Add --ci option to run_vmaf.

* Add --ci option to ffmpeg2vmaf; add tests.

* Add --ci optin to run_vmaf_in_batch; add test.

* Set up --ci option to vmafossexec arguments; to add logic in VmafRunner::run.

* Refactor vmaf.h and vmaf.cpp.

* Refactor vmaf.h/.cpp: move load model functions to LibsvmNusvrTrainTestModel.

* Refactor vmaf.h/.cpp: extract out VmafRunner::_normalize_predict_denormalize().

* Scaffold BootstrapLibsvmNusvrTrainTestModel; remove unnecessary arguments from VmafRunner::_normalize_predict_denormalize().

* Make model polymorphism work.

* Read number of models in bootstrapLibsvmNusvrTrainTestModel; update validation of model type.

* Add virtual destructor to LibsvmNusvrTrainTestModel.

* Make _loadModel static.

* Make svm_model_ptr protected; load svm models to new field bootstrap_svm_model_ptrs.

* Refactor n_subsample.

* Move vmafossexec tests into one class in quality_runner_test.

* Fix test_run_vmafossexec_runner_with_motion2 issue.

* Refactor re n_subsample.

* Refactor _normalize_predict_denormalize: remove subsampling awareness.

* Refactor: extract VmafRunner::_populate_and_normalize_nodes_at_frm method.

* Refactor: extract VmafRunner::_denormalize_prediction method.

* Refactor: extract _transform_score() and _clip_score() method.

* Misc.

* Move _denormalize_prediction from VmafRunner to LibsvmNuvrTrainTestModel.

* Move populate_and_normalize_nodes_at_frm from VmafRunner to LibsvmNusvrTrainTestModel.

* Misc.

* Remove conf_interval option from VmafRunner::run.

* Rename VmafRunner to VmafQualityRunner to make consistent with Python code; Scaffold BootstrapVmafQualityRunner.

* Misc.

* Update predict() to make return a map intead of double; fix type check bug.

* Remove numModels as class attribute in BootstrapLibsvmNusvrTrainTestModel.

* Misc.

* Change bootstrap_svm_model_ptrs from list to vector; add BootstrapLibsvmNusvrTrainTestModel::predict.

* Update _normalize_predict_denormalize_transform_clip to output predictionMaps instead of vmaf.

* Update _transform_score and _clip_score to take predictionMap.

* Refactor.

* Extract _set_prediction_result method.

* Add virtual classes for BootstrapVmafQualityRunner; having bagging scores in log now.

* Add bagging_vmaf to VmafossExecQualityRunner; bagging score mismatch.

* Add stddev to vmafossexec --ci output.

* Misc.

* Add virtual destructor.

* Add ci95_low and ci95_high field to vmafossexec --ci output; update vmafoss doc version.

* Fix _get_model_i_filename issue.

* Simplify predict function by returning map instead map reference.

* Update README.md and CHANGELOG.md.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant