# TorchMetrics Extension
[](https://pypi.org/project/torchmetrics-ext/)
[](https://badge.fury.io/py/torchmetrics-ext)
[](https://github.com/eamonn-zh/torchmetrics_ext/blob/master/LICENSE)
## Installation
Simple installation from PyPI
```bash
pip install torchmetrics-ext
```
## What is TorchMetrics Extension
It is an extension of [torchmetrics](https://lightning.ai/docs/torchmetrics/) containing more metrics for machine learning tasks. Currently, it offers metrics for:
- 3D Visual Grounding
- [ScanRefer](https://daveredrum.github.io/ScanRefer/)
- [ReferIt3D](https://referit3d.github.io/)
## Using TorchMetrics Extension
Here are examples for using the metrics in TorchMetrics Extension:
### ScanRefer
It measures the thresholded accuracy Acc@kIoU, where the positive predictions have higher intersection over union (IoU) with the ground truths than the thresholds. The metric is based on the [ScanRefer](https://daveredrum.github.io/ScanRefer/).
```python
import torch
from torchmetrics_ext.visual_grounding import ScanReferMetric
metric = ScanReferMetric()
# min max bounds of 3D axis-aligned bounding boxes (B, 2, 3)
pred_aabbs = torch.tensor([[[0., 0., 0.], [1., 1., 1.]], [[0., 0., 0.], [2., 2., 2.]]], dtype=torch.float32)
gt_aabbs = torch.tensor([[[0., 0., 0.], [1., 1., 1.]], [[0., 0., 0.], [1.5, 1.5, 1.5]]], dtype=torch.float32)
gt_eval_types = ("unique", "multiple")
results = metric(pred_aabbs, gt_aabbs, gt_eval_types)
```
### ReferIt3D
It measures the accuracy of selecting the target object from the candidates. The metric is based on the [ReferIt3D](https://referit3d.github.io/).
```python
import torch
from torchmetrics_ext.visual_grounding import ReferIt3DMetric
metric = ReferIt3DMetric()
# indices of predicted and ground truth objects (B, )
pred_indices = torch.tensor([5, 2, 0, 0], dtype=torch.uint8)
gt_indices = torch.tensor([5, 5, 1, 0], dtype=torch.uint8)
gt_eval_types = (("easy", "view_dep"), ("easy", "view_indep"), ("hard", "view_dep"), ("hard", "view_dep"))
results = metric(pred_indices, gt_indices, gt_eval_types)
```
Raw data
{
"_id": null,
"home_page": "https://github.com/eamonn-zh/torchmetrics_ext",
"name": "torchmetrics-ext",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "deep learning, machine learning, pytorch, metrics",
"author": "Yiming Zhang",
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/90/d5/7f2bcecd4a4bcbfa36ef93ed13c419442de4836ab36c4c257fa68cbb77dc/torchmetrics_ext-0.2.1.tar.gz",
"platform": null,
"description": "# TorchMetrics Extension\n[](https://pypi.org/project/torchmetrics-ext/)\n[](https://badge.fury.io/py/torchmetrics-ext)\n[](https://github.com/eamonn-zh/torchmetrics_ext/blob/master/LICENSE)\n\n## Installation\nSimple installation from PyPI\n```bash\npip install torchmetrics-ext\n```\n\n## What is TorchMetrics Extension\nIt is an extension of [torchmetrics](https://lightning.ai/docs/torchmetrics/) containing more metrics for machine learning tasks. Currently, it offers metrics for:\n- 3D Visual Grounding\n - [ScanRefer](https://daveredrum.github.io/ScanRefer/)\n - [ReferIt3D](https://referit3d.github.io/)\n\n## Using TorchMetrics Extension\nHere are examples for using the metrics in TorchMetrics Extension:\n\n### ScanRefer\nIt measures the thresholded accuracy Acc@kIoU, where the positive predictions have higher intersection over union (IoU) with the ground truths than the thresholds. The metric is based on the [ScanRefer](https://daveredrum.github.io/ScanRefer/).\n```python\nimport torch\nfrom torchmetrics_ext.visual_grounding import ScanReferMetric\nmetric = ScanReferMetric()\n\n# min max bounds of 3D axis-aligned bounding boxes (B, 2, 3)\npred_aabbs = torch.tensor([[[0., 0., 0.], [1., 1., 1.]], [[0., 0., 0.], [2., 2., 2.]]], dtype=torch.float32)\ngt_aabbs = torch.tensor([[[0., 0., 0.], [1., 1., 1.]], [[0., 0., 0.], [1.5, 1.5, 1.5]]], dtype=torch.float32)\n\ngt_eval_types = (\"unique\", \"multiple\")\nresults = metric(pred_aabbs, gt_aabbs, gt_eval_types)\n```\n\n### ReferIt3D\nIt measures the accuracy of selecting the target object from the candidates. The metric is based on the [ReferIt3D](https://referit3d.github.io/).\n```python\nimport torch\nfrom torchmetrics_ext.visual_grounding import ReferIt3DMetric\nmetric = ReferIt3DMetric()\n\n# indices of predicted and ground truth objects (B, )\npred_indices = torch.tensor([5, 2, 0, 0], dtype=torch.uint8)\ngt_indices = torch.tensor([5, 5, 1, 0], dtype=torch.uint8)\n\ngt_eval_types = ((\"easy\", \"view_dep\"), (\"easy\", \"view_indep\"), (\"hard\", \"view_dep\"), (\"hard\", \"view_dep\"))\nresults = metric(pred_indices, gt_indices, gt_eval_types)\n```\n",
"bugtrack_url": null,
"license": null,
"summary": "An extension of torchmetrics package.",
"version": "0.2.1",
"project_urls": {
"Homepage": "https://github.com/eamonn-zh/torchmetrics_ext"
},
"split_keywords": [
"deep learning",
" machine learning",
" pytorch",
" metrics"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "d006ffad332238162fb31042c4d1a75dbe7a280e8bf14d94f920b6ff921afb36",
"md5": "46794c59570b946acabc84a314b536ac",
"sha256": "34fdf592d34d655a49d66bf34a87ff6f7e607e95788e61da0a3cacb56b0e4031"
},
"downloads": -1,
"filename": "torchmetrics_ext-0.2.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "46794c59570b946acabc84a314b536ac",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 12844,
"upload_time": "2024-10-17T22:06:02",
"upload_time_iso_8601": "2024-10-17T22:06:02.394434Z",
"url": "https://files.pythonhosted.org/packages/d0/06/ffad332238162fb31042c4d1a75dbe7a280e8bf14d94f920b6ff921afb36/torchmetrics_ext-0.2.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "90d57f2bcecd4a4bcbfa36ef93ed13c419442de4836ab36c4c257fa68cbb77dc",
"md5": "3199f112656b049d2620d0903e6f0c90",
"sha256": "04486dc30b2037074e797c2ce046cb3487dfd671fad1647bbc95173ae974cd50"
},
"downloads": -1,
"filename": "torchmetrics_ext-0.2.1.tar.gz",
"has_sig": false,
"md5_digest": "3199f112656b049d2620d0903e6f0c90",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 11112,
"upload_time": "2024-10-17T22:06:03",
"upload_time_iso_8601": "2024-10-17T22:06:03.675297Z",
"url": "https://files.pythonhosted.org/packages/90/d5/7f2bcecd4a4bcbfa36ef93ed13c419442de4836ab36c4c257fa68cbb77dc/torchmetrics_ext-0.2.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-10-17 22:06:03",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "eamonn-zh",
"github_project": "torchmetrics_ext",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "torchmetrics-ext"
}
JSON
