修改 lanms 包安装的判定条件 #740
修改 lanms 包安装的判定条件 #740
Conversation
|
|
||
| def test_lanms(self): | ||
| expect_result = np.array([[8.611, 10, 8.611, 56.11, 30, 56.11, 30, 10, 1.8], [0, 0, 0, 20, 10, 20, 10, 0, 0.8]]) | ||
| assert np.allclose(merge_quadrangle_n9([box1, box2, box3]), expect_result, 1e-2) | ||
| assert np.allclose(merge_quadrangle_n9([box1, box2, box3]), expect_result, 1e-2) is True |
There was a problem hiding this comment.
np.allclose 可直接返回比较结果，无需使用 is True
There was a problem hiding this comment.
np.allclose 可直接返回比较结果，无需使用 is True
是的，之后统一清理小问题。
| @@ -16,8 +28,15 @@ def test_weighted_merge(self): | |||
| assert np.allclose(weighted_merge(box1, box2), expect_result, rtol=1e-2) is True | |||
|
|
|||
| def test_standard_nms(self): | |||
| assert np.allclose(standard_nms([box2, box3], 0.5), box3, 1e-5) | |||
| assert np.allclose(standard_nms([box2, box3], 0.5), box3, 1e-5) is True | |||
|
|
|||
| def test_lanms(self): | |||
There was a problem hiding this comment.
希望测试方法名清晰地描述它们的功能，如test_standard_nms、test_calculate_iou描述都是清晰的，但 test_lanms 稍显笼统，不便明确含义，建议可以更具体一些（如 test_merge_quadrangle_n9_with_multiple_boxes）
There was a problem hiding this comment.
希望测试方法名清晰地描述它们的功能，如test_standard_nms、test_calculate_iou描述都是清晰的，但 test_lanms 稍显笼统，不便明确含义，建议可以更具体一些（如 test_merge_quadrangle_n9_with_multiple_boxes）
lanms 是完整的算法。这个是 E2E 的测试。
| for origin_box_test in origin_boxes_test: | ||
| real_results.append(sorted(merge_quadrangle_n9(origin_box_test), key=lambda x: x[0])) | ||
| for i, real_result in enumerate(real_results): | ||
| assert np.allclose(real_result, expect_processed_boxes_test[i], 1e-2) is True |
There was a problem hiding this comment.
建议添加断言错误的提示消息，以便在断言失败时方便问题定位
There was a problem hiding this comment.
建议添加断言错误的提示消息，以便在断言失败时方便问题定位
测试内容在函数名称已经写明。
| @@ -16,8 +28,15 @@ def test_weighted_merge(self): | |||
| assert np.allclose(weighted_merge(box1, box2), expect_result, rtol=1e-2) is True | |||
|
|
|||
| def test_standard_nms(self): | |||
| assert np.allclose(standard_nms([box2, box3], 0.5), box3, 1e-5) | |||
| assert np.allclose(standard_nms([box2, box3], 0.5), box3, 1e-5) is True | |||
|
|
|||
There was a problem hiding this comment.
建议添加断言错误的提示消息，以便在断言失败时方便问题定位
There was a problem hiding this comment.
建议添加断言错误的提示消息，以便在断言失败时方便问题定位
同上
| data = json.loads(f.readline()) | ||
| origin_boxes_test.append(np.array(data["origin_boxes"])) | ||
| expect_processed_boxes_test.append(sorted(np.array(data["processed_boxes"]), key=lambda x: x[0])) | ||
|
|
||
|
|
||
| class TestLanmsPy: |
There was a problem hiding this comment.
建议为类以及函数添加适当的文档，解释具体的功能与参数
There was a problem hiding this comment.
建议为类以及函数添加适当的文档，解释具体的功能与参数
本部分代码较短，变量名称已经覆盖意义，多余的注释意义不大。
当以下 2 个条件均满足时，才引入 lanms 包。
（1）python 版本 <= 3.10
（2）系统为 linux
Thank you for your contribution to the MindOCR repo.
Before submitting this PR, please make sure:
Motivation
(Write your motivation for proposed changes here.)
Test Plan
python 3.9 + Ascend 910B + Linux 下验证
Related Issues and PRs
(Is this PR part of a group of changes? Link the other relevant PRs and Issues here. Use https://help.github.com/en/articles/closing-issues-using-keywords for help on GitHub syntax)