GithubHelp home page GithubHelp logo

Comments (5)

Sp2-Hybrid avatar Sp2-Hybrid commented on August 22, 2024

你好,我想问一下,你按照作者的原始配置文件跑起来正常吗?
我使用作者原始配置文件,文件为configs/dota/s2anet_r50_fpn_1x.py,我在train的过程中发现loss会变成nan,将lr改小之后loss来回波动,最终也会变成nan。。。非常费解。。。

from s2anet.

yzk-lab avatar yzk-lab commented on August 22, 2024

你好,我想问一下,你按照作者的原始配置文件跑起来正常吗?
我使用作者原始配置文件,文件为configs/dota/s2anet_r50_fpn_1x.py,我在train的过程中发现loss会变成nan,将lr改小之后loss来回波动,最终也会变成nan。。。非常费解。。。

正常 就是测试结果有问题,loss一直在下降

from s2anet.

csuhan avatar csuhan commented on August 22, 2024

I guess your checkpoints is not well trained or configured.

As shown in test.py, (1) please check the output log of each categories during testing. (2) visualize the results to see whether your checkpoint is 'correct'. Also, you can download the checkpoints provided by us to verify the results.

s2anet/tools/test.py

Lines 223 to 236 in 173ef21

elif data_name == 'dota':
dst_path = osp.dirname(args.out)
dst_raw_path = osp.join(dst_path, 'result_raw')
dst_merge_path = osp.join(dst_path, 'result_merge')
if not osp.exists(dst_path):
os.mkdir(dst_path)
if not osp.exists(dst_raw_path):
os.mkdir(dst_raw_path)
if not osp.exists(dst_merge_path):
os.mkdir(dst_merge_path)
print('convert result to dota result format at {}'.format(dst_raw_path))
result2dota_task1(outputs, dst_raw_path, dataset)
print('merge result')
mergebypoly(dst_raw_path, dst_merge_path)

from s2anet.

yzk-lab avatar yzk-lab commented on August 22, 2024

谢谢你的回复,我已经解决了这个问题并且取得了较高的map.现在我有个新问题,就是如果我要训练自己的数据集(800x600的jpg文件和DOTA的png不太一样),我数据集标签是8个点的坐标的形式。我转成VOC格式可以做吗?因为我不想使用prepare_dota.py去处理数据集了。

from s2anet.

csuhan avatar csuhan commented on August 22, 2024

跟annotation的原始格式没有关系。但模型在训练时需要回归bbox的角度,因此在preprocessing之后必须转成(x,y,w,h,a)的形式。
如果原始数据是8点的polygon,则需要自定义一个dataset,读入8参数,输出5参数的形式,因此最终的数据格式和dota_obb应该是一样的。

from s2anet.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.