faster-coco-eval 1.7.1__tar.gz → 1.8.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- faster_coco_eval-1.8.0/MANIFEST.in +5 -0
- faster_coco_eval-1.8.0/PKG-INFO +278 -0
- faster_coco_eval-1.8.0/README.md +234 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/coco_eval/cocoeval.cpp +388 -140
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/coco_eval/dataset.cpp +6 -5
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/coco_eval/dataset.h +2 -2
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/mask_api/src/mask.cpp +212 -90
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/mask_api/src/mask.h +32 -2
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/mask_api/src/rle.cpp +34 -10
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/core/coco.py +126 -84
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/core/cocoeval.py +111 -81
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/core/faster_eval_api.py +198 -39
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/core/mask.py +12 -19
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/curves.py +40 -21
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/display.py +14 -13
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/draw.py +49 -34
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/extra.py +21 -22
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/utils.py +4 -1
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/utils/pytorch/coco_dataset.py +7 -10
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/utils/pytorch/coco_eval.py +19 -22
- faster_coco_eval-1.8.0/faster_coco_eval/version.py +2 -0
- faster_coco_eval-1.8.0/faster_coco_eval.egg-info/PKG-INFO +278 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval.egg-info/SOURCES.txt +14 -1
- faster_coco_eval-1.8.0/history.md +228 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/pyproject.toml +16 -4
- faster_coco_eval-1.8.0/requirements/docs.txt +6 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/setup.py +5 -6
- faster_coco_eval-1.8.0/tests/test_annotation_id_zero.py +56 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_basic.py +151 -1
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_boundary.py +5 -1
- faster_coco_eval-1.8.0/tests/test_coco_api_and_evaluator_regressions.py +562 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_coco_metric.py +3 -1
- faster_coco_eval-1.8.0/tests/test_compute_oks.py +69 -0
- faster_coco_eval-1.8.0/tests/test_core_low_regressions.py +105 -0
- faster_coco_eval-1.8.0/tests/test_cpp_safety_regressions.py +108 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_crowdpose.py +4 -1
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_dataset.py +226 -0
- faster_coco_eval-1.8.0/tests/test_display_matrix_normalization.py +42 -0
- faster_coco_eval-1.8.0/tests/test_extensive_pycocotools_comparison.py +589 -0
- faster_coco_eval-1.8.0/tests/test_extra_curves.py +58 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_extra_draw.py +36 -8
- faster_coco_eval-1.8.0/tests/test_extra_regressions.py +92 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_extra_utils.py +42 -6
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_keypoints.py +14 -4
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_lvis_metric.py +2 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_mask_api.py +249 -0
- faster_coco_eval-1.8.0/tests/test_orphan_annotations.py +88 -0
- faster_coco_eval-1.8.0/tests/test_pytorch_coco_dataset.py +56 -0
- faster_coco_eval-1.8.0/tests/test_pytorch_coco_eval.py +68 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_simple_extra.py +146 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_torchmetrics.py +19 -8
- faster_coco_eval-1.8.0/tests/test_world_evaluator.py +124 -0
- faster_coco_eval-1.7.1/MANIFEST.in +0 -1
- faster_coco_eval-1.7.1/PKG-INFO +0 -189
- faster_coco_eval-1.7.1/README.md +0 -142
- faster_coco_eval-1.7.1/faster_coco_eval/version.py +0 -2
- faster_coco_eval-1.7.1/faster_coco_eval.egg-info/PKG-INFO +0 -189
- faster_coco_eval-1.7.1/requirements/docs.txt +0 -8
- faster_coco_eval-1.7.1/tests/test_world_evalutor.py +0 -84
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/LICENSE +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/coco_eval/cocoeval.h +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/coco_eval/types.h +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/faster_eval_api/faster_eval_api.cpp +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/csrc/mask_api/mask_api.cpp +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/__init__.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/core/__init__.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/extra/__init__.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/utils/__init__.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval/utils/pytorch/__init__.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval.egg-info/dependency_links.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval.egg-info/requires.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/faster_coco_eval.egg-info/top_level.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/requirements/extra.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/requirements/runtime.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/requirements/tests.txt +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/setup.cfg +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_cocoapi_fake_data.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_init_pycocotools.py +0 -0
- {faster_coco_eval-1.7.1 → faster_coco_eval-1.8.0}/tests/test_ranges.py +0 -0
|
@@ -0,0 +1,278 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: faster-coco-eval
|
|
3
|
+
Version: 1.8.0
|
|
4
|
+
Summary: Faster interpretation of the original COCOEval
|
|
5
|
+
Home-page: https://github.com/MiXaiLL76/faster_coco_eval
|
|
6
|
+
Author: MiXaiLL76,Borda
|
|
7
|
+
Author-email: mike.milos@yandex.ru
|
|
8
|
+
Classifier: Development Status :: 4 - Beta
|
|
9
|
+
Classifier: Operating System :: OS Independent
|
|
10
|
+
Classifier: Programming Language :: Python :: 3
|
|
11
|
+
Classifier: Programming Language :: Python :: 3.10
|
|
12
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
13
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
14
|
+
Classifier: Programming Language :: Python :: 3.13
|
|
15
|
+
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
|
|
16
|
+
Requires-Python: >=3.10
|
|
17
|
+
Description-Content-Type: text/markdown
|
|
18
|
+
License-File: LICENSE
|
|
19
|
+
Requires-Dist: numpy
|
|
20
|
+
Provides-Extra: extra
|
|
21
|
+
Requires-Dist: numpy; extra == "extra"
|
|
22
|
+
Requires-Dist: opencv-python-headless; extra == "extra"
|
|
23
|
+
Requires-Dist: pandas; extra == "extra"
|
|
24
|
+
Requires-Dist: plotly; extra == "extra"
|
|
25
|
+
Requires-Dist: Pillow; extra == "extra"
|
|
26
|
+
Provides-Extra: tests
|
|
27
|
+
Requires-Dist: numpy; extra == "tests"
|
|
28
|
+
Requires-Dist: numpy; extra == "tests"
|
|
29
|
+
Requires-Dist: pycocotools; python_version < "3.13" and extra == "tests"
|
|
30
|
+
Requires-Dist: pytest; extra == "tests"
|
|
31
|
+
Requires-Dist: pytest-cov; extra == "tests"
|
|
32
|
+
Requires-Dist: parameterized; extra == "tests"
|
|
33
|
+
Dynamic: author
|
|
34
|
+
Dynamic: author-email
|
|
35
|
+
Dynamic: classifier
|
|
36
|
+
Dynamic: description
|
|
37
|
+
Dynamic: description-content-type
|
|
38
|
+
Dynamic: home-page
|
|
39
|
+
Dynamic: license-file
|
|
40
|
+
Dynamic: provides-extra
|
|
41
|
+
Dynamic: requires-dist
|
|
42
|
+
Dynamic: requires-python
|
|
43
|
+
Dynamic: summary
|
|
44
|
+
|
|
45
|
+
# Faster-COCO-Eval
|
|
46
|
+
|
|
47
|
+
**The Fastest, Most Reliable COCO Evaluation Library for Computer Vision**
|
|
48
|
+
|
|
49
|
+
[](https://pypi.org/project/faster-coco-eval)
|
|
50
|
+
[](https://pypi.org/project/faster-coco-eval/)
|
|
51
|
+
[](https://anaconda.org/conda-forge/faster-coco-eval)
|
|
52
|
+
[](https://anaconda.org/conda-forge/faster-coco-eval)
|
|
53
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/wiki)
|
|
54
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/LICENSE)
|
|
55
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/actions/workflows/unittest.yml)
|
|
56
|
+
|
|
57
|
+
## Replace pycocotools with Faster-COCO-Eval Today
|
|
58
|
+
|
|
59
|
+
| Aspect | pycocotools | **faster-coco-eval** |
|
|
60
|
+
| ------------------------------ | ------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
|
|
61
|
+
| **Support & Development** | Outdated and not actively maintained. Issues and incompatibilities arise with new releases. | Actively maintained, continuously evolving, and regularly updated with new features and bug fixes. |
|
|
62
|
+
| **Transparency & Reliability** | Lacks comprehensive testing, making updates risky and results less predictable. | Emphasizes extensive test coverage and code quality, ensuring trustworthy and reliable results. |
|
|
63
|
+
| **Performance** | Significantly slower, especially on large datasets or distributed workloads. | Native C++ implementation with measured speedups that depend on the workload and hardware. |
|
|
64
|
+
| **Functionality** | Limited to basic COCO format evaluation. | Offers extended metrics, support for new IoU types, compatibility with more datasets (e.g., CrowdPose, LVIS), advanced visualizations, and seamless integration with PyTorch/TorchVision. |
|
|
65
|
+
| **Ease of Use** | Requires manual installation, often with compilation issues. | Simple `pip install` with pre-built wheels for common CPython/glibc Linux, macOS, and Windows platforms; source builds may be needed elsewhere. |
|
|
66
|
+
| **Visualization** | Basic plotting capabilities. | Advanced error visualization, annotation display, and comprehensive metric analysis tools. |
|
|
67
|
+
|
|
68
|
+
______________________________________________________________________
|
|
69
|
+
|
|
70
|
+
**Key Benefits of Faster-COCO-Eval:**
|
|
71
|
+
|
|
72
|
+
✅ **Blazing Fast Performance** - Evaluate large datasets in minutes instead of hours
|
|
73
|
+
✅ **Reliable & Trusted** - Extensive test coverage ensures consistent, reproducible results
|
|
74
|
+
✅ **Modern Features** - Support for latest CV tasks, IoU types, and dataset formats
|
|
75
|
+
✅ **Easy to Use** - Drop-in replacement for pycocotools with enhanced API
|
|
76
|
+
✅ **Comprehensive Visualization** - Understand your model's performance with beautiful, informative plots
|
|
77
|
+
|
|
78
|
+
**Join thousands of computer vision researchers and engineers who have already switched to Faster-COCO-Eval!**
|
|
79
|
+
|
|
80
|
+
## Quick Installation
|
|
81
|
+
|
|
82
|
+
### Option 1: Basic (Drop-in Replacement)
|
|
83
|
+
|
|
84
|
+
Get started in seconds with the core evaluation functionality:
|
|
85
|
+
|
|
86
|
+
```bash
|
|
87
|
+
pip install faster-coco-eval
|
|
88
|
+
```
|
|
89
|
+
|
|
90
|
+
### Option 2: Full Installation (with Visualization)
|
|
91
|
+
|
|
92
|
+
For complete functionality including advanced visualization tools:
|
|
93
|
+
|
|
94
|
+
```bash
|
|
95
|
+
pip install faster-coco-eval[extra]
|
|
96
|
+
```
|
|
97
|
+
|
|
98
|
+
### Option 3: Conda Installation
|
|
99
|
+
|
|
100
|
+
If you use Anaconda/Miniconda:
|
|
101
|
+
|
|
102
|
+
```bash
|
|
103
|
+
conda install conda-forge::faster-coco-eval
|
|
104
|
+
```
|
|
105
|
+
|
|
106
|
+
## 🚀 Quick Start: Drop-in Replacement
|
|
107
|
+
|
|
108
|
+
Replace pycocotools with Faster-COCO-Eval in **2 lines of code**:
|
|
109
|
+
|
|
110
|
+
```python
|
|
111
|
+
import faster_coco_eval
|
|
112
|
+
|
|
113
|
+
# This single line replaces pycocotools with faster-coco-eval
|
|
114
|
+
faster_coco_eval.init_as_pycocotools()
|
|
115
|
+
|
|
116
|
+
# Now use the familiar pycocotools API
|
|
117
|
+
from pycocotools.coco import COCO
|
|
118
|
+
from pycocotools.cocoeval import COCOeval
|
|
119
|
+
|
|
120
|
+
# Load annotations and predictions
|
|
121
|
+
anno_json = "path/to/annotations.json"
|
|
122
|
+
pred_json = "path/to/predictions.json"
|
|
123
|
+
anno = COCO(str(anno_json)) # Annotations file
|
|
124
|
+
pred = anno.loadRes(str(pred_json)) # Predictions file
|
|
125
|
+
|
|
126
|
+
# Evaluate bounding boxes
|
|
127
|
+
val = COCOeval(anno, pred, "bbox")
|
|
128
|
+
val.evaluate()
|
|
129
|
+
val.accumulate()
|
|
130
|
+
val.summarize()
|
|
131
|
+
|
|
132
|
+
# Or evaluate segmentation masks
|
|
133
|
+
val = COCOeval(anno, pred, "segm")
|
|
134
|
+
val.evaluate()
|
|
135
|
+
val.accumulate()
|
|
136
|
+
val.summarize()
|
|
137
|
+
```
|
|
138
|
+
|
|
139
|
+
**That's it! Your existing code runs through the native evaluator with no API changes.**
|
|
140
|
+
|
|
141
|
+
## ⚡ Blazing Fast Performance
|
|
142
|
+
|
|
143
|
+
Faster-COCO-Eval uses a native C++ implementation. The measured result below is one local reference point; real-world speedup varies with dataset size, metric, and hardware.
|
|
144
|
+
|
|
145
|
+
### Real-World Performance Benchmark
|
|
146
|
+
|
|
147
|
+
Local reference measurement: 100 synthetic images, 15 ground-truth boxes and 100 detections per image, five categories, bbox evaluation, seven timed samples after two warmups. The measurement was run on 2026-08-23 on macOS 26.6.1 arm64, Python 3.10.11, and NumPy 2.2.6. The implementations' `eval["precision"]` arrays matched within the comparison tolerance.
|
|
148
|
+
|
|
149
|
+
| Evaluation Type | Faster-COCO-Eval (sec) | pycocotools (sec) | Speedup |
|
|
150
|
+
| --------------- | ----------------------- | ----------------------- | --------- |
|
|
151
|
+
| Bounding Boxes | 0.053977 (MAD 0.001883) | 0.296960 (MAD 0.001667) | **5.50x** |
|
|
152
|
+
|
|
153
|
+
This local measurement is hardware- and workload-dependent; it is not a universal performance guarantee.
|
|
154
|
+
|
|
155
|
+
### Colab Examples
|
|
156
|
+
|
|
157
|
+
See the performance in action:
|
|
158
|
+
|
|
159
|
+
- [mmdetection comparison](https://nbviewer.org/github/MiXaiLL76/faster_coco_eval/blob/main/examples/comparison/mmdet/colab_example.ipynb)
|
|
160
|
+
- [ultralytics comparison](https://nbviewer.org/github/MiXaiLL76/faster_coco_eval/blob/main/examples/comparison/ultralytics/colab_example.ipynb)
|
|
161
|
+
|
|
162
|
+
## 🎯 Powerful Features
|
|
163
|
+
|
|
164
|
+
Faster-COCO-Eval goes beyond basic evaluation with these advanced capabilities:
|
|
165
|
+
|
|
166
|
+
### Core Evaluation
|
|
167
|
+
|
|
168
|
+
- **Drop-in pycocotools replacement** - No code changes needed
|
|
169
|
+
- **Support for all COCO metric types**: bbox, segm, keypoints
|
|
170
|
+
- **LVIS (Large Vocabulary Instance Segmentation) evaluation**
|
|
171
|
+
- **CrowdPose and custom keypoint datasets**
|
|
172
|
+
- **Multiple IoU types**: standard, rotated, and custom IoU definitions
|
|
173
|
+
|
|
174
|
+
### Advanced Visualization
|
|
175
|
+
|
|
176
|
+
- **Error visualization**: See where your model is making mistakes
|
|
177
|
+
- **Annotation display**: Visualize ground truth and predictions together
|
|
178
|
+
- **Metric curves**: Precision-recall curves, class-wise performance
|
|
179
|
+
- **Confusion matrices and error analysis**
|
|
180
|
+
- **Interactive Jupyter notebook examples**
|
|
181
|
+
|
|
182
|
+
### Modern Integrations
|
|
183
|
+
|
|
184
|
+
- **PyTorch/TorchVision compatibility**
|
|
185
|
+
- **Seamless integration with mmdetection and YOLO frameworks**
|
|
186
|
+
- **Distributed evaluation support**
|
|
187
|
+
- **Memory optimized for large datasets**
|
|
188
|
+
|
|
189
|
+
### Additional Tools
|
|
190
|
+
|
|
191
|
+
- **Boundary evaluation for segmentation tasks**
|
|
192
|
+
- **Custom dataset support**
|
|
193
|
+
- **Comprehensive API documentation**
|
|
194
|
+
- **Extensive test coverage and reliability**
|
|
195
|
+
|
|
196
|
+
## ✅ Testing & Reliability
|
|
197
|
+
|
|
198
|
+
Faster-COCO-Eval prioritizes **correctness and reliability** through extensive testing:
|
|
199
|
+
|
|
200
|
+
### Comprehensive Test Suite
|
|
201
|
+
|
|
202
|
+
- **90+ automated tests** covering all functionality
|
|
203
|
+
- **Numerical parity checks** against pycocotools across the supported metrics
|
|
204
|
+
- **Continuous integration** on Python 3.10-3.13
|
|
205
|
+
- **Edge case coverage** including boundary conditions and error handling
|
|
206
|
+
|
|
207
|
+
### Extensive PyCocoTools Comparison
|
|
208
|
+
|
|
209
|
+
New comprehensive tests validate **numerical parity within a strict tolerance** with pycocotools:
|
|
210
|
+
|
|
211
|
+
- **Object Detection**: Tests with 10-100 images, hundreds to thousands of annotations
|
|
212
|
+
- **Instance Segmentation**: RLE mask encoding and pixel-level IoU validation
|
|
213
|
+
- **Keypoint Detection**: 17-keypoint pose estimation with varied visibility
|
|
214
|
+
- **Multiple Scenarios**: Small/medium/large objects, various confidence distributions
|
|
215
|
+
- **Edge Cases**: Perfect predictions, low-confidence detections, mixed object sizes
|
|
216
|
+
|
|
217
|
+
The comparison suite checks numerical parity between faster_coco_eval and pycocotools, giving you confidence to use this library as a drop-in replacement. See the benchmark above for one measured bbox workload; it is not a universal performance guarantee.
|
|
218
|
+
|
|
219
|
+
See [tests/README.md](tests/README.md) for detailed test documentation.
|
|
220
|
+
|
|
221
|
+
## 📚 Comprehensive Documentation
|
|
222
|
+
|
|
223
|
+
### Usage Examples
|
|
224
|
+
|
|
225
|
+
Explore practical, runnable examples in Jupyter notebooks:
|
|
226
|
+
|
|
227
|
+
- [Basic Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/eval_example.html) - Get started with COCO evaluation
|
|
228
|
+
- [Metric Curves](https://mixaill76.github.io/faster_coco_eval/examples/curve_example.html) - Precision-recall and metric visualization
|
|
229
|
+
- [LVIS Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/lvis_example.html) - Large vocabulary instance segmentation
|
|
230
|
+
- [CrowdPose Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/crowdpose_example.html) - Keypoint detection for crowded scenes
|
|
231
|
+
- [Custom Keypoints](https://mixaill76.github.io/faster_coco_eval/examples/ced_example.html) - Extend to custom keypoint datasets
|
|
232
|
+
- [Annotation Visualization](https://mixaill76.github.io/faster_coco_eval/examples/show_example.html) - Display and analyze annotations
|
|
233
|
+
|
|
234
|
+
### Detailed Documentation
|
|
235
|
+
|
|
236
|
+
- [Official Wiki](https://github.com/MiXaiLL76/faster_coco_eval/wiki) - Complete API reference and guides
|
|
237
|
+
- [Changelog](https://mixaill76.github.io/faster_coco_eval/history.html) - Latest updates and improvements
|
|
238
|
+
- [API Documentation](https://github.com/MiXaiLL76/faster_coco_eval/wiki) - Detailed function documentation
|
|
239
|
+
|
|
240
|
+
## ⭐ Star History
|
|
241
|
+
|
|
242
|
+
[](https://star-history.com/#MiXaiLL76/faster_coco_eval&Date)
|
|
243
|
+
|
|
244
|
+
## 📄 License
|
|
245
|
+
|
|
246
|
+
Faster-COCO-Eval is distributed under the Apache 2.0 license. See [LICENSE](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/LICENSE) for more information.
|
|
247
|
+
|
|
248
|
+
## 📚 Citation
|
|
249
|
+
|
|
250
|
+
If you use Faster-COCO-Eval in your research, please cite:
|
|
251
|
+
|
|
252
|
+
```bibtex
|
|
253
|
+
@article{faster-coco-eval,
|
|
254
|
+
title = {{Faster-COCO-Eval}: Faster and Enhanced COCO Evaluation Library},
|
|
255
|
+
author = {MiXaiLL76},
|
|
256
|
+
year = {2024}
|
|
257
|
+
}
|
|
258
|
+
```
|
|
259
|
+
|
|
260
|
+
## 🤝 Contributing
|
|
261
|
+
|
|
262
|
+
We welcome contributions! Check out our [CONTRIBUTING.md](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/CONTRIBUTING.md) for guidelines on how to get started.
|
|
263
|
+
|
|
264
|
+
## 🐛 Issues and Support
|
|
265
|
+
|
|
266
|
+
If you encounter any issues or have questions:
|
|
267
|
+
|
|
268
|
+
1. Check the [Wiki](https://github.com/MiXaiLL76/faster_coco_eval/wiki) for common solutions
|
|
269
|
+
2. Search existing [issues](https://github.com/MiXaiLL76/faster_coco_eval/issues)
|
|
270
|
+
3. Open a new issue with detailed information about your problem
|
|
271
|
+
|
|
272
|
+
## 🚀 Get Started Today
|
|
273
|
+
|
|
274
|
+
```bash
|
|
275
|
+
pip install faster-coco-eval[extra]
|
|
276
|
+
```
|
|
277
|
+
|
|
278
|
+
**Replace pycocotools with Faster-COCO-Eval and experience evaluation at lightning speed!**
|
|
@@ -0,0 +1,234 @@
|
|
|
1
|
+
# Faster-COCO-Eval
|
|
2
|
+
|
|
3
|
+
**The Fastest, Most Reliable COCO Evaluation Library for Computer Vision**
|
|
4
|
+
|
|
5
|
+
[](https://pypi.org/project/faster-coco-eval)
|
|
6
|
+
[](https://pypi.org/project/faster-coco-eval/)
|
|
7
|
+
[](https://anaconda.org/conda-forge/faster-coco-eval)
|
|
8
|
+
[](https://anaconda.org/conda-forge/faster-coco-eval)
|
|
9
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/wiki)
|
|
10
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/LICENSE)
|
|
11
|
+
[](https://github.com/MiXaiLL76/faster_coco_eval/actions/workflows/unittest.yml)
|
|
12
|
+
|
|
13
|
+
## Replace pycocotools with Faster-COCO-Eval Today
|
|
14
|
+
|
|
15
|
+
| Aspect | pycocotools | **faster-coco-eval** |
|
|
16
|
+
| ------------------------------ | ------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
|
|
17
|
+
| **Support & Development** | Outdated and not actively maintained. Issues and incompatibilities arise with new releases. | Actively maintained, continuously evolving, and regularly updated with new features and bug fixes. |
|
|
18
|
+
| **Transparency & Reliability** | Lacks comprehensive testing, making updates risky and results less predictable. | Emphasizes extensive test coverage and code quality, ensuring trustworthy and reliable results. |
|
|
19
|
+
| **Performance** | Significantly slower, especially on large datasets or distributed workloads. | Native C++ implementation with measured speedups that depend on the workload and hardware. |
|
|
20
|
+
| **Functionality** | Limited to basic COCO format evaluation. | Offers extended metrics, support for new IoU types, compatibility with more datasets (e.g., CrowdPose, LVIS), advanced visualizations, and seamless integration with PyTorch/TorchVision. |
|
|
21
|
+
| **Ease of Use** | Requires manual installation, often with compilation issues. | Simple `pip install` with pre-built wheels for common CPython/glibc Linux, macOS, and Windows platforms; source builds may be needed elsewhere. |
|
|
22
|
+
| **Visualization** | Basic plotting capabilities. | Advanced error visualization, annotation display, and comprehensive metric analysis tools. |
|
|
23
|
+
|
|
24
|
+
______________________________________________________________________
|
|
25
|
+
|
|
26
|
+
**Key Benefits of Faster-COCO-Eval:**
|
|
27
|
+
|
|
28
|
+
✅ **Blazing Fast Performance** - Evaluate large datasets in minutes instead of hours
|
|
29
|
+
✅ **Reliable & Trusted** - Extensive test coverage ensures consistent, reproducible results
|
|
30
|
+
✅ **Modern Features** - Support for latest CV tasks, IoU types, and dataset formats
|
|
31
|
+
✅ **Easy to Use** - Drop-in replacement for pycocotools with enhanced API
|
|
32
|
+
✅ **Comprehensive Visualization** - Understand your model's performance with beautiful, informative plots
|
|
33
|
+
|
|
34
|
+
**Join thousands of computer vision researchers and engineers who have already switched to Faster-COCO-Eval!**
|
|
35
|
+
|
|
36
|
+
## Quick Installation
|
|
37
|
+
|
|
38
|
+
### Option 1: Basic (Drop-in Replacement)
|
|
39
|
+
|
|
40
|
+
Get started in seconds with the core evaluation functionality:
|
|
41
|
+
|
|
42
|
+
```bash
|
|
43
|
+
pip install faster-coco-eval
|
|
44
|
+
```
|
|
45
|
+
|
|
46
|
+
### Option 2: Full Installation (with Visualization)
|
|
47
|
+
|
|
48
|
+
For complete functionality including advanced visualization tools:
|
|
49
|
+
|
|
50
|
+
```bash
|
|
51
|
+
pip install faster-coco-eval[extra]
|
|
52
|
+
```
|
|
53
|
+
|
|
54
|
+
### Option 3: Conda Installation
|
|
55
|
+
|
|
56
|
+
If you use Anaconda/Miniconda:
|
|
57
|
+
|
|
58
|
+
```bash
|
|
59
|
+
conda install conda-forge::faster-coco-eval
|
|
60
|
+
```
|
|
61
|
+
|
|
62
|
+
## 🚀 Quick Start: Drop-in Replacement
|
|
63
|
+
|
|
64
|
+
Replace pycocotools with Faster-COCO-Eval in **2 lines of code**:
|
|
65
|
+
|
|
66
|
+
```python
|
|
67
|
+
import faster_coco_eval
|
|
68
|
+
|
|
69
|
+
# This single line replaces pycocotools with faster-coco-eval
|
|
70
|
+
faster_coco_eval.init_as_pycocotools()
|
|
71
|
+
|
|
72
|
+
# Now use the familiar pycocotools API
|
|
73
|
+
from pycocotools.coco import COCO
|
|
74
|
+
from pycocotools.cocoeval import COCOeval
|
|
75
|
+
|
|
76
|
+
# Load annotations and predictions
|
|
77
|
+
anno_json = "path/to/annotations.json"
|
|
78
|
+
pred_json = "path/to/predictions.json"
|
|
79
|
+
anno = COCO(str(anno_json)) # Annotations file
|
|
80
|
+
pred = anno.loadRes(str(pred_json)) # Predictions file
|
|
81
|
+
|
|
82
|
+
# Evaluate bounding boxes
|
|
83
|
+
val = COCOeval(anno, pred, "bbox")
|
|
84
|
+
val.evaluate()
|
|
85
|
+
val.accumulate()
|
|
86
|
+
val.summarize()
|
|
87
|
+
|
|
88
|
+
# Or evaluate segmentation masks
|
|
89
|
+
val = COCOeval(anno, pred, "segm")
|
|
90
|
+
val.evaluate()
|
|
91
|
+
val.accumulate()
|
|
92
|
+
val.summarize()
|
|
93
|
+
```
|
|
94
|
+
|
|
95
|
+
**That's it! Your existing code runs through the native evaluator with no API changes.**
|
|
96
|
+
|
|
97
|
+
## ⚡ Blazing Fast Performance
|
|
98
|
+
|
|
99
|
+
Faster-COCO-Eval uses a native C++ implementation. The measured result below is one local reference point; real-world speedup varies with dataset size, metric, and hardware.
|
|
100
|
+
|
|
101
|
+
### Real-World Performance Benchmark
|
|
102
|
+
|
|
103
|
+
Local reference measurement: 100 synthetic images, 15 ground-truth boxes and 100 detections per image, five categories, bbox evaluation, seven timed samples after two warmups. The measurement was run on 2026-08-23 on macOS 26.6.1 arm64, Python 3.10.11, and NumPy 2.2.6. The implementations' `eval["precision"]` arrays matched within the comparison tolerance.
|
|
104
|
+
|
|
105
|
+
| Evaluation Type | Faster-COCO-Eval (sec) | pycocotools (sec) | Speedup |
|
|
106
|
+
| --------------- | ----------------------- | ----------------------- | --------- |
|
|
107
|
+
| Bounding Boxes | 0.053977 (MAD 0.001883) | 0.296960 (MAD 0.001667) | **5.50x** |
|
|
108
|
+
|
|
109
|
+
This local measurement is hardware- and workload-dependent; it is not a universal performance guarantee.
|
|
110
|
+
|
|
111
|
+
### Colab Examples
|
|
112
|
+
|
|
113
|
+
See the performance in action:
|
|
114
|
+
|
|
115
|
+
- [mmdetection comparison](https://nbviewer.org/github/MiXaiLL76/faster_coco_eval/blob/main/examples/comparison/mmdet/colab_example.ipynb)
|
|
116
|
+
- [ultralytics comparison](https://nbviewer.org/github/MiXaiLL76/faster_coco_eval/blob/main/examples/comparison/ultralytics/colab_example.ipynb)
|
|
117
|
+
|
|
118
|
+
## 🎯 Powerful Features
|
|
119
|
+
|
|
120
|
+
Faster-COCO-Eval goes beyond basic evaluation with these advanced capabilities:
|
|
121
|
+
|
|
122
|
+
### Core Evaluation
|
|
123
|
+
|
|
124
|
+
- **Drop-in pycocotools replacement** - No code changes needed
|
|
125
|
+
- **Support for all COCO metric types**: bbox, segm, keypoints
|
|
126
|
+
- **LVIS (Large Vocabulary Instance Segmentation) evaluation**
|
|
127
|
+
- **CrowdPose and custom keypoint datasets**
|
|
128
|
+
- **Multiple IoU types**: standard, rotated, and custom IoU definitions
|
|
129
|
+
|
|
130
|
+
### Advanced Visualization
|
|
131
|
+
|
|
132
|
+
- **Error visualization**: See where your model is making mistakes
|
|
133
|
+
- **Annotation display**: Visualize ground truth and predictions together
|
|
134
|
+
- **Metric curves**: Precision-recall curves, class-wise performance
|
|
135
|
+
- **Confusion matrices and error analysis**
|
|
136
|
+
- **Interactive Jupyter notebook examples**
|
|
137
|
+
|
|
138
|
+
### Modern Integrations
|
|
139
|
+
|
|
140
|
+
- **PyTorch/TorchVision compatibility**
|
|
141
|
+
- **Seamless integration with mmdetection and YOLO frameworks**
|
|
142
|
+
- **Distributed evaluation support**
|
|
143
|
+
- **Memory optimized for large datasets**
|
|
144
|
+
|
|
145
|
+
### Additional Tools
|
|
146
|
+
|
|
147
|
+
- **Boundary evaluation for segmentation tasks**
|
|
148
|
+
- **Custom dataset support**
|
|
149
|
+
- **Comprehensive API documentation**
|
|
150
|
+
- **Extensive test coverage and reliability**
|
|
151
|
+
|
|
152
|
+
## ✅ Testing & Reliability
|
|
153
|
+
|
|
154
|
+
Faster-COCO-Eval prioritizes **correctness and reliability** through extensive testing:
|
|
155
|
+
|
|
156
|
+
### Comprehensive Test Suite
|
|
157
|
+
|
|
158
|
+
- **90+ automated tests** covering all functionality
|
|
159
|
+
- **Numerical parity checks** against pycocotools across the supported metrics
|
|
160
|
+
- **Continuous integration** on Python 3.10-3.13
|
|
161
|
+
- **Edge case coverage** including boundary conditions and error handling
|
|
162
|
+
|
|
163
|
+
### Extensive PyCocoTools Comparison
|
|
164
|
+
|
|
165
|
+
New comprehensive tests validate **numerical parity within a strict tolerance** with pycocotools:
|
|
166
|
+
|
|
167
|
+
- **Object Detection**: Tests with 10-100 images, hundreds to thousands of annotations
|
|
168
|
+
- **Instance Segmentation**: RLE mask encoding and pixel-level IoU validation
|
|
169
|
+
- **Keypoint Detection**: 17-keypoint pose estimation with varied visibility
|
|
170
|
+
- **Multiple Scenarios**: Small/medium/large objects, various confidence distributions
|
|
171
|
+
- **Edge Cases**: Perfect predictions, low-confidence detections, mixed object sizes
|
|
172
|
+
|
|
173
|
+
The comparison suite checks numerical parity between faster_coco_eval and pycocotools, giving you confidence to use this library as a drop-in replacement. See the benchmark above for one measured bbox workload; it is not a universal performance guarantee.
|
|
174
|
+
|
|
175
|
+
See [tests/README.md](tests/README.md) for detailed test documentation.
|
|
176
|
+
|
|
177
|
+
## 📚 Comprehensive Documentation
|
|
178
|
+
|
|
179
|
+
### Usage Examples
|
|
180
|
+
|
|
181
|
+
Explore practical, runnable examples in Jupyter notebooks:
|
|
182
|
+
|
|
183
|
+
- [Basic Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/eval_example.html) - Get started with COCO evaluation
|
|
184
|
+
- [Metric Curves](https://mixaill76.github.io/faster_coco_eval/examples/curve_example.html) - Precision-recall and metric visualization
|
|
185
|
+
- [LVIS Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/lvis_example.html) - Large vocabulary instance segmentation
|
|
186
|
+
- [CrowdPose Evaluation](https://mixaill76.github.io/faster_coco_eval/examples/crowdpose_example.html) - Keypoint detection for crowded scenes
|
|
187
|
+
- [Custom Keypoints](https://mixaill76.github.io/faster_coco_eval/examples/ced_example.html) - Extend to custom keypoint datasets
|
|
188
|
+
- [Annotation Visualization](https://mixaill76.github.io/faster_coco_eval/examples/show_example.html) - Display and analyze annotations
|
|
189
|
+
|
|
190
|
+
### Detailed Documentation
|
|
191
|
+
|
|
192
|
+
- [Official Wiki](https://github.com/MiXaiLL76/faster_coco_eval/wiki) - Complete API reference and guides
|
|
193
|
+
- [Changelog](https://mixaill76.github.io/faster_coco_eval/history.html) - Latest updates and improvements
|
|
194
|
+
- [API Documentation](https://github.com/MiXaiLL76/faster_coco_eval/wiki) - Detailed function documentation
|
|
195
|
+
|
|
196
|
+
## ⭐ Star History
|
|
197
|
+
|
|
198
|
+
[](https://star-history.com/#MiXaiLL76/faster_coco_eval&Date)
|
|
199
|
+
|
|
200
|
+
## 📄 License
|
|
201
|
+
|
|
202
|
+
Faster-COCO-Eval is distributed under the Apache 2.0 license. See [LICENSE](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/LICENSE) for more information.
|
|
203
|
+
|
|
204
|
+
## 📚 Citation
|
|
205
|
+
|
|
206
|
+
If you use Faster-COCO-Eval in your research, please cite:
|
|
207
|
+
|
|
208
|
+
```bibtex
|
|
209
|
+
@article{faster-coco-eval,
|
|
210
|
+
title = {{Faster-COCO-Eval}: Faster and Enhanced COCO Evaluation Library},
|
|
211
|
+
author = {MiXaiLL76},
|
|
212
|
+
year = {2024}
|
|
213
|
+
}
|
|
214
|
+
```
|
|
215
|
+
|
|
216
|
+
## 🤝 Contributing
|
|
217
|
+
|
|
218
|
+
We welcome contributions! Check out our [CONTRIBUTING.md](https://github.com/MiXaiLL76/faster_coco_eval/blob/main/CONTRIBUTING.md) for guidelines on how to get started.
|
|
219
|
+
|
|
220
|
+
## 🐛 Issues and Support
|
|
221
|
+
|
|
222
|
+
If you encounter any issues or have questions:
|
|
223
|
+
|
|
224
|
+
1. Check the [Wiki](https://github.com/MiXaiLL76/faster_coco_eval/wiki) for common solutions
|
|
225
|
+
2. Search existing [issues](https://github.com/MiXaiLL76/faster_coco_eval/issues)
|
|
226
|
+
3. Open a new issue with detailed information about your problem
|
|
227
|
+
|
|
228
|
+
## 🚀 Get Started Today
|
|
229
|
+
|
|
230
|
+
```bash
|
|
231
|
+
pip install faster-coco-eval[extra]
|
|
232
|
+
```
|
|
233
|
+
|
|
234
|
+
**Replace pycocotools with Faster-COCO-Eval and experience evaluation at lightning speed!**
|