Jason Lee

Results 36 comments of Jason Lee

@dsandii @liuheng92 input_key = 'input_images:0';output_key = 'resnet_v1_50/conv1/BatchNorm/moving_mean:0' I think maybe these two are input and output and successfully save the pb file, but i stucked at the following code when...

def main_eval_1(argv=None): import os os.environ['CUDA_VISIBLE_DEVICES'] = FLAGS.gpu_list try: os.makedirs(FLAGS.output_dir) except OSError as e: if e.errno != 17: raise from tensorflow.python.saved_model import tag_constants signature_key = tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY input_key = 'input_images:0';output_key = 'resnet_v1_50/conv1/BatchNorm/moving_mean:0'...

八个数字后面加文本string,因为是训练检测模型,这个string我写的是null(随便写都可以),可以正常训练。 On Tue, Nov 19, 2019 at 2:11 PM Asuna88 wrote: > > 这个模型我down下来跑过,里面的多进程会导致内存溢出,每次step内存持续变大,初步估计自动回收机制无法实现实时回收,手动gc也失效。我将generator去掉,单进程跑,不会出现溢出,目前还没完美的解决办法 > > 你好,,我是这么处理多线程转单线程的,不知道对否,大神请指点。 > 在data_util.py文件里面,注释掉下面几句话,然后再补充了三句话,请问大神这样做对否。 > ( 请问如果写成 workers=1, max_queue_size =2 , 这个是表示什么意思呢, ) > (好像线程thread的增加是这句话self._threads.append(thread)...

@wsy915 你好,你在什么数据上面训练对,最终loss到了多少,我对loss最后知道0.19

I have met the same problem, can @sjvasquez @el3ment you help me? Thank you for your time.

@Duankaiwen I have trained one category object detection centernet based on your project but I found that in test.py my model can only get low scores(about 0.4-0.6). It confused me...

your GPU memory is limited.

set batch size to 1 and chunk_size [1]