We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
task1：人车25类，task2:烟火5类 最后两个epoch的评测分别是：
person [00]: AP50=18.907, #pred=11432 car [01]: AP50=49.915, #pred=17987 pickup truck [02]: AP50=12.394, #pred=153 minibus [03]: AP50=26.414, #pred=1832 motorhome [04]: AP50=0.000, #pred=0 truck [05]: AP50=41.789, #pred=10869 forklift [06]: AP50=2.242, #pred=82 pushdozer [07]: AP50=24.122, #pred=889 crane [08]: AP50=34.870, #pred=291 road roller [09]: AP50=12.273, #pred=77 excavator [10]: AP50=44.692, #pred=952 mixer truck [11]: AP50=35.062, #pred=640 oil tank truck [12]: AP50=15.678, #pred=247 watering cart [13]: AP50=18.945, #pred=300 mixer truck [14]: AP50=0.000, #pred=0 fire fighting truck [15]: AP50=22.007, #pred=102 garbage truck [16]: AP50=0.000, #pred=62 bus [17]: AP50=31.796, #pred=429 tractors [18]: AP50=10.476, #pred=9 bicycle [19]: AP50=37.750, #pred=419 electric scooter [20]: AP50=0.000, #pred=2 motorbike [21]: AP50=37.470, #pred=2551 tricycle [22]: AP50=19.092, #pred=1410 train [23]: AP50=70.461, #pred=123 other [24]: AP50=4.537, #pred=2470 smoke [25]: AP50=11.684, #pred=4877 fire [26]: AP50=4.240, #pred=1942 cloud [27]: AP50=1.390, #pred=53 fog [28]: AP50=0.000, #pred=195 chimney [29]: AP50=0.451, #pred=435 unknown [30]: AR50=0.000, #pred=344874 01/17 00:57:02 - mmengine - INFO - known classes has 60830 predictions. 01/17 00:57:02 - mmengine - INFO - unknown classes has 344874 predictions. 01/17 00:57:02 - mmengine - INFO - Wilderness Impact: {50: 0.0} 01/17 00:57:02 - mmengine - INFO - Absolute OSE (total_num_unk_det_as_known): {50: 0.0} 01/17 00:57:02 - mmengine - INFO - total_num_unk 0 01/17 00:57:02 - mmengine - INFO - Prev class AP50: 22.835738700029136 01/17 00:57:02 - mmengine - INFO - Prev class Precisions50: 17.36332553942016 01/17 00:57:02 - mmengine - INFO - Prev class Recall50: 40.26247657567848 01/17 00:57:02 - mmengine - INFO - Current class AP50: 3.552976728791721 01/17 00:57:02 - mmengine - INFO - Current class Precisions50: 9.496200795228047 01/17 00:57:02 - mmengine - INFO - Current class Recall50: 12.623674063553134 01/17 00:57:02 - mmengine - INFO - Known AP50: 19.621945038156237 01/17 00:57:02 - mmengine - INFO - Known Precisions50: 16.052138082054807 01/17 00:57:02 - mmengine - INFO - Known Recall50: 35.65600949032425 01/17 00:57:02 - mmengine - INFO - Unknown AP50: 0.0 01/17 00:57:02 - mmengine - INFO - Unknown Precisions50: 0.0 01/17 00:57:02 - mmengine - INFO - Unknown Recall50: 0.0
person [00]: AP50=18.907, #pred=11432 car [01]: AP50=49.915, #pred=17987 pickup truck [02]: AP50=12.394, #pred=153 minibus [03]: AP50=26.414, #pred=1832 motorhome [04]: AP50=0.000, #pred=0 truck [05]: AP50=41.789, #pred=10869 forklift [06]: AP50=2.242, #pred=82 pushdozer [07]: AP50=24.122, #pred=889 crane [08]: AP50=34.870, #pred=291 road roller [09]: AP50=12.273, #pred=77 excavator [10]: AP50=44.692, #pred=952 mixer truck [11]: AP50=35.062, #pred=640 oil tank truck [12]: AP50=15.678, #pred=247 watering cart [13]: AP50=18.945, #pred=300 mixer truck [14]: AP50=0.000, #pred=0 fire fighting truck [15]: AP50=22.007, #pred=102 garbage truck [16]: AP50=0.000, #pred=62 bus [17]: AP50=31.796, #pred=429 tractors [18]: AP50=10.476, #pred=9 bicycle [19]: AP50=37.750, #pred=419 electric scooter [20]: AP50=0.000, #pred=2 motorbike [21]: AP50=37.470, #pred=2551 tricycle [22]: AP50=19.092, #pred=1410 train [23]: AP50=70.461, #pred=123 other [24]: AP50=4.537, #pred=2470 smoke [25]: AP50=11.672, #pred=4916 fire [26]: AP50=4.230, #pred=1961 cloud [27]: AP50=1.353, #pred=43 fog [28]: AP50=0.000, #pred=182 chimney [29]: AP50=0.426, #pred=444 unknown [30]: AR50=0.000, #pred=345066 01/17 00:58:17 - mmengine - INFO - known classes has 60874 predictions. 01/17 00:58:17 - mmengine - INFO - unknown classes has 345066 predictions. 01/17 00:58:17 - mmengine - INFO - Wilderness Impact: {50: 0.0} 01/17 00:58:17 - mmengine - INFO - Absolute OSE (total_num_unk_det_as_known): {50: 0.0} 01/17 00:58:17 - mmengine - INFO - total_num_unk 0 01/17 00:58:17 - mmengine - INFO - Prev class AP50: 22.835738700029136 01/17 00:58:17 - mmengine - INFO - Prev class Precisions50: 17.36332553942016 01/17 00:58:17 - mmengine - INFO - Prev class Recall50: 40.26247657567848 01/17 00:58:17 - mmengine - INFO - Current class AP50: 3.5363268206818597 01/17 00:58:17 - mmengine - INFO - Current class Precisions50: 9.955457272587786 01/17 00:58:17 - mmengine - INFO - Current class Recall50: 12.634267283836007 01/17 00:58:17 - mmengine - INFO - Known AP50: 19.619170053471258 01/17 00:58:17 - mmengine - INFO - Known Precisions50: 16.12868082828143 01/17 00:58:17 - mmengine - INFO - Known Recall50: 35.65777502703807 01/17 00:58:17 - mmengine - INFO - Unknown AP50: 0.0 01/17 00:58:17 - mmengine - INFO - Unknown Precisions50: 0.0 01/17 00:58:17 - mmengine - INFO - Unknown Recall50: 0.0
task1的指标没有变，task2的指标变化， 我看代码里应该是使用task1训练的best模型评测task1，在task1的best基础上用task2数据训练，用task2的best模型评测task2。 但是我用task2的best模型去测试task1的数据，发现task1上的表现比直接用task1的best上模型测的效果差了很多，在task1的map上下降了9个点，R下降不多，P下降的很多，说明用task2数据微调后的模型对task1影响很大，我可视化两次结果，发现目标还是检测的到（R变化不大），主要是用task2微调后，目标的class错误率提升了（P下降多）
The text was updated successfully, but these errors were encountered:
No branches or pull requests
task1：人车25类，task2:烟火5类
最后两个epoch的评测分别是：
task1的指标没有变，task2的指标变化，
我看代码里应该是使用task1训练的best模型评测task1，在task1的best基础上用task2数据训练，用task2的best模型评测task2。
但是我用task2的best模型去测试task1的数据，发现task1上的表现比直接用task1的best上模型测的效果差了很多，在task1的map上下降了9个点，R下降不多，P下降的很多，说明用task2数据微调后的模型对task1影响很大，我可视化两次结果，发现目标还是检测的到（R变化不大），主要是用task2微调后，目标的class错误率提升了（P下降多）
The text was updated successfully, but these errors were encountered: