History
v1.6.7
[x] Added workflow for automatic Sphinx documentation build and deploy (
.github/workflows/docs.yml
)[x] Improved main build workflow (
build.yml
): now includes documentation generation task and enhanced handling of environment variables and package publishing[x] Updated testing workflow (
test.yaml
): added support for multiple Python versions and triggering via workflow_call[x] Major refactoring and optimization of the C++ core (
cocoeval.cpp
,cocoeval.h
, newdataset.cpp
): improved readability, redesigned data structures, and optimized evaluation logic for better speed[x] Significant improvements to annotation parsing, serialization, and evaluation handling in the C++ part
[x] Added new hooks and formatters (e.g., clang-format) to
.pre-commit-config.yaml
, cleaned up and structured existing sections[x] Updated README.md: added comparison table between pycocotools and faster-coco-eval, improved project advantages section
[x] Performed code decomposition, optimization, and standardization of variables in workflow files
[x] Modern C++ idioms applied throughout the codebase for better readability and performance
v1.6.6
[x] Support pytorch dist FasterCocoEvaluator
[x] Support torchvision CocoDetection Dataset as FasterCocoDetection
[x] Return COCO.showAnns https://github.com/MiXaiLL76/faster_coco_eval/issues/52
v1.6.5
[x] Support CrowdPose Dataset
[x] add CrowdPose simple test
[x] ImageEvaluation pickleable https://github.com/MiXaiLL76/faster_coco_eval/issues/45
[x] Create test for MP eval https://github.com/MiXaiLL76/faster_coco_eval/blob/main/tests/test_world_evalutor.py
v1.6.4
[x] fix windows mask api.
[x] numpy>=2 support for py39+
[x] update CI\CD for build macos & windows whl
v1.6.1 - v1.6.3
[x] fix windows & macos build for torchmetrics
v1.6.0
[x] Rework mask_api with pybind11 C++ .
[x] Rework RLE support.
[x] Create test files for all components.
[x] The math_matches function has been reworked, with an emphasis on using C++ code.
[x] Added more documentation of functions. Optimized existing ones.
[x] Added rleToBoundary func with 2 backend [“mask_api”, “opencv”]
[x] IoU type boundary support (further testing is needed)
[x] Create async rle and boundary comput discussion
v1.5.7
[x] Compare COCOEval bug fix.
v1.5.6
[x] Replace CED MSE curve with MAE (px) curve
[x] Add CED examples
[x] Display IoU and MAE for keypoints
[x] Reworked eval._prepare to clear up the return flow
[x] Reworked the C++ part of COCOevalEvaluateImages and COCOevalAccumulate
[x] Add new COCOevalEvaluateAccumulate to combine these two calls. You can use old style separate_eval==True (default=False)
[x] COCOevalAccumulate & COCOevalEvaluateAccumulate -> COCOeval_faster.eval is now correctly created as numpy arrays.
[x] Append LVIS dataset support lvis_style=True in COCOeval_faster
cocoEval = COCOeval_faster(cocoGt, cocoDt, iouType, lvis_style=True, print_function=print)
cocoEval.params.maxDets = [300]
v1.5.5
[x] Add CED MSE curve
[x] Review tests
[x] Review COCOeval_faster.math_matches function and COCOeval_faster.compute_mIoU function
v1.5.3 - v1.5.4
[x] Worked out the ability to work with skeletons and various key points
[x]
eval.state_as_dict
Now works for key points
v1.5.2
[x] Change comparison to colab_example
[x] append utils with opencv conver_mask_to_poly (extra)
[x] append drop_cocodt_by_score for extra eval
v1.5.1
[x] breaking change | new static function COCO.load_json
[x] new curve f1_confidence with
cur.plot_f1_confidence()
[x] breaking change | replace display_matrix arg
in_percent
tonormalize
[x] breaking change | rework draw functions
v1.4.2
[x] append Auto-formatters
[x] append py36 support
[x] append pandas to requirements for plotly[express]
[x] update mask api with pycootools
v1.4.1
[x] append Plotly fig return
[x] append preview GT only func. Without eval.
cocoGt = COCO(...)
preview = PreviewResults(cocoGt, iouType='segm')
preview.display_tp_fp_fn(data_folder=..., image_ids=..., display_gt=True)
v1.4.0
[x] fix issue https://github.com/MiXaiLL76/faster_coco_eval/issues/12
[x] Updated pre-rec calculation method
[x] Updated required libraries
[x] Moved all matplotlib dependencies to plotly
[x] Append new examples & mmeval test file
v1.3.3
[x] fix by ViTrox https://github.com/vitrox-technologies/faster_coco_eval
missing file issue
issue discovered by torchmetric
fstring for python3.7
Windows compilation
v1.3.2
[x] rework math_matches function. moved to faster_eval_api
[x] Moved calculations from python to c++
[x] Separated extra classes
[x] Added new sample data
[x] append mIoU based on TP pred.
[x] append mAUC based on Coco pre/rec.
v1.3.1
[x] rework mask code
[x] change np.float to float (numpy deprecations)
v1.3.0
[x] remove pycocotools dependencies
[x] clean c/c++ code
v1.2.3
[x] Implemented of mean IoU for TP
[x] set FP-red FN-blue
v1.2.2
[x] Removed own implementation of pre-rec
[x] Switched to the implementation of pre-rec calculation from COCO eval
[x] Lost backward compatibility
[x] Implemented output fp/fn/tp + gt to pictures
v1.2.1
[x] bug fix with pre-rec curve
[x] rework error calc (tp/fp/fn)
[x] change image plot to plotly
[x] append docker auto builder
[x] append native iou calc (slow but accurate)
[x] rework auc calc with link
v1.1.3-v1.1.4
[x] rebuild plotly backend
[x] Segm bug-fix
v1.1.2
[x] Append fp fn error analysis via curves
[x] Append confusion matrix
[x] Append plotly backend support for ROC / AUC
v1.1.1
[x] Redesigned curves
[x] Reworked data preload
[x] Append csrc to setup
[x] Build sdist Package
v1.1.0
[x] Wrap c++ code
[x] Get it to compile
[x] Add COCOEval class wraper
[x] Remove detectron2 dependencies
[x] Remove torch dependencies
[x] Append unittest
[x] Append ROC / AUC curves
[x] Check if it works on windows