Yuze Shi

I found two problems here. The first one is when there are paths in compile_commands.json, ncm-clang will remove all the "\\\\" and which is auto generated by cmake when in...

readme 里面 **Train** 部分有提到要改标注文件名称或者代码 改代码的话就把 **icdar.py** 里 604 行改成 txt_fn = im_fn.replace(os.path.basename(im_fn), 'gt_%s.txt' % os.path.basename(im_fn).split('.')[0]) 就好了 Note: you should change the gt text file of icdar2015's filename to img_*.txt instead...

> > > > > > > @dvlee1024，请问下这个测试代码是哪个，我训练出来的模型yolov3.data-00000-of-00001这个，image_demo里面是yolo3.weight，这个是怎么转的或者其他方式测试吗，谢谢！ > > 请问这个问题解决了吗？方便介绍一下是怎么转换成.weight的吗？ test.py 48 行有个 `model.load_weights("./yolov3")` 可以直接加载训练保存的权重

> > > > @bright-zy select toStartOfHour(event_time) as t, countIf(event_type = 1) as i, sumIf(num_source_parts , event_type = 2) as m from cnch(server, system.server_part_log) where event_date >= today() - 1...

> 看日志有报错： PartitionSelector: Failed to load server_part_log digest information from system.server_part_log > > 目前通过设置：alter table measure_olap.rpt_databus_task_run_detail_di_cube_v3 modify setting cnch_merge_round_robin_partitions_interval = 180, max_partition_for_multi_select = 10; 解决 > > server_part_log 加载失败可能是什么导致的？ 这个问题近期内部版本有个修复，刚看了下...