Huangqqqhhh

python setup.py build develop

1

报错 段错误 (核心已转储)

IC15

1

请问一下ic15的test.json在哪下载，之前发的好像是错的，检测不到文本

rapidfuzz里面找不到string_metric，导致在pretrain测试的时候中断

4

[09/02 19:07:09] d2.evaluation.evaluator INFO: Total inference time: 0:08:48.902549 (3.647604 s / img per device, on 2 devices) [09/02 19:07:09] d2.evaluation.evaluator INFO: Total inference pure compute time: 0:01:17 (0.536733 s /...

batch_size

8

REC_HEAD: BATCH_SIZE: 128 SOLVER: IMS_PER_BATCH: 8 IMS_PER_BATCH是什么

请问一下，在跑demo的时候遇到一下问题，应该如何解决呢？ [06/27 10:30:24 detectron2]: Arguments: Namespace(confidence_threshold=0.3, config_file='/home/hhq/data/code/DeepSolo/configs/R_50/IC15/finetune_150k_tt_mlt_13_15.yaml', input=['/home/hhq/data/code/DeepSolo/imputimg/IMG_20230626_204410.jpg'], opts=['MODEL.WEIGHTS', '/home/hhq/data/code/DeepSolo/weight/ic15_res50_finetune_synth-tt-mlt-13-15.pth'], output='/home/hhq/data/code/DeepSolo/outputimg', video_input=None, webcam=False) 段错误 (核心已转储)

Thanks for your excellent work, is Detectron2 downloaded into the glass environment as a packet?

RuntimeError: cannot reshape tensor of 0 elements into shape [0, 256, -1] because the unspecified dimension size -1 can be any value and is ambiguous 经常碰到这个问题，虽然后面能继续跑，但是可能会影响效果，请问该怎么样避免。

请问检测batch size调整为2 识别batch size调整为32，学习率按比例调整为原来四分之一，迭代次数不变，发现训练的效果很差，所以请问这样调整合适嘛？