Aurora11111

@OKhyc 是的啊

reference:https://github.com/Aurora11111/text-detection-train-yolov3 在 2019-11-01 16:16:16，"zlszhonglongshen" 写道： @Aurora11111 你上面提供crnn非常好，我想问下，该如何训练文字检测呢？ — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

@mona-alz reference: https://github.com/Aurora11111/speaker-recognition-pytorch

@nidhal1231 you can create it also by python dvector_create.py . I kown you want to use them in uis-rnn, but test_sequence.npy, test_cluster_ids.npy there are used to test single speaker-id.if you...

@GauravWaghmare how to set the epoch when I trianing about 300 speaker,and how much speech per person need?

refference: https://github.com/Aurora11111/synthdata-zh https://github.com/Aurora11111/TextRecognitionDataGenerator

@那个alphabets里面包含有英文和中文的

@yeah I trained.

@Zhangpeixiang I haven't meet this problem I think you can reduce your batch size or resized your image size smaller.

@Zhangpeixiang maybe you shoule resize the test picture to match with the size of your trianed model's picture