-
Sound_event_detection中跑baseline时,run.py line 111 为
model = getattr(models, config['model']["type"])(
inputdim=trainloader.dataset.datadim,
outputdim=len(label_to_idx),
…
-
-
crnn怎么训练的,是要自己写一个训练代码吗
fryng updated
6 years ago
-
我在训练模型的时候,使用gpu进行训练,但是还没开始输出loss数据,训练机的内存(16G)就全部给吃掉了,最后报“OSError: [Errno 12] Cannot allocate memory”的错误,请问训练这个模型的时候需要用多大的内存呢?
-
对app.py进行了修改,用于视频中的文字检测与识别。对于一些视频,运行一段时间后,出现如下错误:
File "main_chineseocr.py", line 31, in TEXT
POST(img)
File "main_chineseocr.py", line 49, in POST
leftAdjust=True,rightAdjust=True,alp…
-
![default](https://user-images.githubusercontent.com/22487109/46000351-e314e100-c0da-11e8-9ab2-7ebbb69ecd09.png)
-
请问作者,你的ncnn模型是怎么手动优化的,你的模型我在安卓手机可以调用,我自己训练的psenet的模型也已经转成了ncnn的param和bin格式的模型但是不能调用,麻烦能不能指点一下 @ouyanghuiyu
-
CRNN的输入是把图片的高度放缩到32,如果按照相同的比例将宽放缩,此时不同的图片的宽度是不一样的,这种情况应该如何训练模型?
-
train_loader = torch.utils.data.DataLoader(
traindataset, batch_size=batchSize,
shuffle=False, sampler=None,
num_workers=int(workers),
collate_fn=alignCollate(imgH=imgH, imgW=imgW,…
-
使用作者的数据集源代码训练的,使用的环境配置为:python3.6+pytorch1.3
中间调了三个bug
1. 由于在'cpu_texts = [clean_txt(tx.decode('utf-8')) for tx in cpu_texts]'这句时会报错str没有decode的方法,查询资料后感觉是python2和python3的区别,便把代码改成了'cpu_texts = [cle…