wentao

非常感谢，我这边再试试！ 祝好！ william

您好！我这边已经对voc2007数据集训练完成，但是发现在测试的时候出现： “{0: '8', 1: '6', 2: '12', 3: '1', 4: '14', 5: '7', 6: '11', 7: '18', 8: '0', 9: '19', 10: '2', 11: '4', 12: '13', 13: '15', 14:...

> os.listdir()是列出文件夹下所有文件夹和文件，看test.py的意思是把所有的待测试图片放在test_base_path下，而不是把路径写到txt里面，这样应该可以做预测。这个项目看起来是不支持直接测试mAP的，你需要自己改改train.py才能测试mAP。另外请教一下，你的训练过程batchsize是如何设定的？对代码进行了大改吗？   此致 颂好 ——————————   > […](#) > ------------------ 原始邮件 ------------------ 发件人: "wentao"

是的，需要你自己写一个脚本将voc数据集label的格式转为该工程需要的格式就好！

> same question? > l run test.py in 1 nvidia titan x pascal card, the part log is following: > ... > xxx.jpg > 480 640 > 300 > person...

hello,have you solved the problem？@arunrathore45

> #8452 > > A prototype, first implement asynchronous scheduling, then by patching GroupCoordinator to implement dp. What is patching group coordinator ? Can I verify the performance of dp...

> It says "zmq.error.ZMQError: Address already in use (addr='tcp://localhost:55555')" Yes, that is indeed the problem. But I encountered another problem. I found that when using 2p1d, one of the p...

> [@tensorflowt](https://github.com/tensorflowt) Hey, can you share the configuration file and the workload so that we can better understand the problem? Thanks! I think I haven't modified the configuration file. ...

> [@tensorflowt](https://github.com/tensorflowt) Thanks for sharing! Currently, when the workload is too heavy (i.e., the decode cannot process all the requests in real time), there will be a known issue that...