Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / bubbliiiing/deeplabv3-plus-pytorch issues and pull requests
#127 - 用预测生成的图片看上去很暗,调用摄像头也跟蒙了一层阴影一样
Issue -
State: open - Opened by RexGloria about 1 month ago
- 1 comment
#126 - 为什么我用这个跑,一模一样的怎么miou跑不到70以上啊,up主是怎么跑到70几,我设置了model_path = "",也是加载主干网络,为什么
Issue -
State: open - Opened by WQJ0623 about 1 month ago
#125 - 为什么第51轮就报错
Issue -
State: open - Opened by lxproot about 1 month ago
- 2 comments
#124 - D:\深度学习\Pytorch深度学习\deeplabv3-plus-pytorch-main\logs\loss_2024_12_07_16_43_48 is not a directory
Issue -
State: open - Opened by Yuchen-hub about 2 months ago
#123 - 网络结构图
Issue -
State: open - Opened by jiaqingxiang111 2 months ago
#122 - 网络结构图
Issue -
State: open - Opened by jiaqingxiang111 2 months ago
#121 - 代码请求
Issue -
State: open - Opened by akang-ai 2 months ago
#120 - 同一个数据,同一处理方式,你的代码里unet跑正常,这个miou不变,只预测背景,为啥
Issue -
State: open - Opened by kan2y 6 months ago
- 1 comment
#119 - 运行predict.py结果不正常
Issue -
State: open - Opened by 78748547 6 months ago
#118 - miou,mPA,Accuracy一直不变
Issue -
State: open - Opened by cainiaocq 7 months ago
- 3 comments
#117 - mobilenet_v2.pth.tar文件被损坏
Issue -
State: open - Opened by wakakkakak 8 months ago
#116 - 更换新的主干
Issue -
State: open - Opened by yzyi2000 8 months ago
#115 - 请问代码中的deeplab_mobilenetv2.pth和deeplab_xception.pth,是整个网络的训练权重还是主干网络的训练权重呢?
Issue -
State: open - Opened by ylj183 9 months ago
#114 - mobileNetV2和Xception哪个效果好
Issue -
State: open - Opened by shen369 9 months ago
#113 - 请问如果训练到epoch=59的时候中断了,在程序哪里修改可以继续从59训练呀?
Issue -
State: open - Opened by ylj183 9 months ago
- 2 comments
#112 - 你好,请问我用的是pytorch进行训练,为什么会报错缺少tensorboard
Issue -
State: open - Opened by QingHuan-qh 9 months ago
#111 - voc2012数据集怎么训练
Issue -
State: open - Opened by zhaozzt 9 months ago
#110 - 代码跑出来的miou比论文中写的miou低很多,有什么解决方法吗?
Issue -
State: open - Opened by ylj183 9 months ago
- 3 comments
#109 - Focal Loss 代码
Issue -
State: open - Opened by pumpbumb 10 months ago
#108 - 训练时间过长
Issue -
State: open - Opened by Dasenyoung 10 months ago
#107 - 请问这报错是什么原因
Issue -
State: open - Opened by tandonghua 10 months ago
#106 - 每一个epoch的训练时间太长了
Issue -
State: open - Opened by xbd1519722757 10 months ago
#105 - loss从train一开始就显示=nan
Issue -
State: open - Opened by wakakkakak 11 months ago
- 4 comments
#104 - 单机多卡分布训练
Issue -
State: open - Opened by BaronDuan about 1 year ago
- 3 comments
#103 - predict输出onnx失败
Issue -
State: open - Opened by Dasenyoung about 1 year ago
- 1 comment
#102 - 作者您好,我想咨询您一下为什么num_classes需要在分类个数上+1,请问是和dataset有关吗,还是网络结构或者损失函数的计算?
Issue -
State: closed - Opened by zfhplaycode about 1 year ago
- 2 comments
#101 - miou不变化
Issue -
State: open - Opened by liuq0123 about 1 year ago
- 6 comments
#100 - 感谢up
Issue -
State: open - Opened by CAM-FSS about 1 year ago
- 1 comment
#99 - imgs, pngs, labels = batch
Issue -
State: open - Opened by 13108268049 about 1 year ago
- 2 comments
#98 - 为什么我只加载预训练backbone训练100个epoch,miou只有65?
Issue -
State: open - Opened by VaeReol about 1 year ago
- 1 comment
#97 - 大佬,我在使用 get_miou时,cpu只有单核在跑,是不是可以优化一下,让所有核心都参与进来啊
Issue -
State: open - Opened by ZJDATY about 1 year ago
- 1 comment
#96 - miou一直下降,f_score一直上升
Issue -
State: closed - Opened by VaeReol about 1 year ago
- 3 comments
#95 - ascii”编解码器无法对0-3位置的字符进行编码
Issue -
State: open - Opened by MayLuoA about 1 year ago
#94 - 损失函数路径问题
Issue -
State: open - Opened by redbird12138 about 1 year ago
- 2 comments
#93 - DeepGlobe数据集
Issue -
State: open - Opened by shen369 about 1 year ago
#92 - 请问训练时为什么会卡在第一个epoch不动呀?
Issue -
State: open - Opened by misaka100001 about 1 year ago
- 9 comments
#91 - C++环境下模型预测
Issue -
State: open - Opened by VPRzhou over 1 year ago
- 3 comments
#90 - 冻结阶段和解冻阶段的epoch设置一样,对训练模型有什么影响
Issue -
State: open - Opened by heng1221 over 1 year ago
- 1 comment
#89 - 调用onnx模型
Issue -
State: open - Opened by guoyangang over 1 year ago
#88 - 你的xception网络最后的Exit flow层里conv3,conv4,conv5都没有BN和relu,请问这是遗漏是还是就是不加bn和relu更好
Issue -
State: open - Opened by liuliuyu618 over 1 year ago
- 2 comments
#87 - 预测时发生的问题
Issue -
State: open - Opened by yuijn456 over 1 year ago
#86 - model convert to TorchScript
Issue -
State: open - Opened by jinyu-118 over 1 year ago
- 1 comment
#85 - 我每个batch都跑了miou,那挑最佳模式是找loss最小的,还是miou最高的。
Issue -
State: open - Opened by ZJDATY over 1 year ago
- 6 comments
#84 - 文件路径问题,用下好的VOC文件夹就没问题,用自己做的数据集一训练就出现这个错误
Issue -
State: open - Opened by 1366332545 over 1 year ago
- 7 comments
#83 - miou95%,但是预测结果差之千里
Issue -
State: open - Opened by bibucher over 1 year ago
- 1 comment
#82 - 测试输入文件报错
Issue -
State: open - Opened by loki-keroro over 1 year ago
#81 - MobileNetV3网络在DeepLabv3+中的使用
Issue -
State: open - Opened by Voyagerlemon over 1 year ago
- 11 comments
#80 - 模型权值载入问题
Issue -
State: open - Opened by Voyagerlemon over 1 year ago
- 1 comment
#79 - backbone部分没有载入是错误的
Issue -
State: open - Opened by zidddd9527 over 1 year ago
- 1 comment
#78 - 关于MIoU卡在50上不去的问题
Issue -
State: open - Opened by Gal1nsoga over 1 year ago
- 13 comments
#77 - 希望up再出几个语义分割算法的实战[respect]
Issue -
State: open - Opened by zidddd9527 over 1 year ago
- 5 comments
#76 - (已看过问题汇总依旧没解决=_=)关于predict.py只显示内存地址的问题
Issue -
State: open - Opened by ghost over 1 year ago
- 3 comments
#75 - 根目录:文件夹路径的问题
Issue -
State: open - Opened by BaronDuan over 1 year ago
- 3 comments
#74 - gbk报错
Issue -
State: open - Opened by BaronDuan over 1 year ago
- 5 comments
#73 - fix np.int error and adapt to numpy 1.20+
Pull Request -
State: open - Opened by jackhanyuan over 1 year ago
#72 - 各位大佬们,谁能分享一下用C++写的后处理过程啊。
Issue -
State: open - Opened by ZJDATY over 1 year ago
- 10 comments
#71 - 大佬,我没动主干网络,改了一点主干以外的东西,然后用您给的预训练权重的时候键值就有点对应不上,这是为什么呢?这样也可以继续训练,就是最后得到的miou会不会变得很差?
Issue -
State: open - Opened by liuliuyu618 over 1 year ago
- 4 comments
#70 - 预训练权重
Issue -
State: open - Opened by cabbage479 over 1 year ago
- 1 comment
#70 - 预训练权重
Issue -
State: open - Opened by cabbage479 over 1 year ago
- 1 comment
#69 - 训练集loss还在下降 ,但是验证集,以及剧烈抖动了,这样咋办呢
Issue -
State: closed - Opened by ZJDATY over 1 year ago
- 1 comment
#69 - 训练集loss还在下降 ,但是验证集,以及剧烈抖动了,这样咋办呢
Issue -
State: closed - Opened by ZJDATY over 1 year ago
- 1 comment
#68 - 第25轮设置的解冻,发现一开始的损失是最低的,请问这是过拟合了嘛?怎么解决好一点。还有用您的预训练权重是不是比官方ImageNet权重要更快收,效果更好?
Issue -
State: open - Opened by liuliuyu618 almost 2 years ago
- 9 comments
#67 - 使用VOC12+SBD数据集需要对标签再处理吗
Issue -
State: open - Opened by 1920230345 almost 2 years ago
#66 - 预训练权重损坏
Issue -
State: open - Opened by Liumomo30 almost 2 years ago
#65 - 博主大大,请问能输出识别物每个像素点的坐标信息吗?
Issue -
State: open - Opened by wswjq almost 2 years ago
- 1 comment
#64 - 如何绘制训练过程中fscore曲线?
Issue -
State: open - Opened by zfhplaycode almost 2 years ago
- 3 comments
#63 - CPU负载太高了,GPU每次只会有个小尖峰
Issue -
State: open - Opened by ZJDATY almost 2 years ago
- 16 comments
#62 - 关于AttributeError的问题
Issue -
State: open - Opened by Gal1nsoga almost 2 years ago
- 4 comments
#61 - 关于heatmap热力图可视化
Issue -
State: open - Opened by Vision-LYF almost 2 years ago
- 1 comment
#60 - 为什么pytorch版本的和tensorflow版本的v3+ miou能相差这么多
Issue -
State: closed - Opened by SoramiAkito almost 2 years ago
- 1 comment
#59 - miou的原理是什么?
Issue -
State: open - Opened by rain0322 almost 2 years ago
#58 - 关于自己的数据集标签的问题
Issue -
State: open - Opened by xiaoyaoqingchun almost 2 years ago
- 2 comments
#57 - 运行到51步,跳GPU运行内存不足。batch已经降到2了。
Issue -
State: open - Opened by rain0322 almost 2 years ago
- 6 comments
#56 - 训练数据量
Issue -
State: open - Opened by rain0322 almost 2 years ago
#55 - 大佬能帮忙看看这个报错吗
Issue -
State: closed - Opened by Shan2L almost 2 years ago
- 2 comments
#54 - 指标计算
Issue -
State: closed - Opened by zhoujian-z almost 2 years ago
#53 - 数据集更换为cityscapes,在训练时报错
Issue -
State: open - Opened by littlesharkchili almost 2 years ago
- 3 comments
#52 - 参数量问题
Issue -
State: open - Opened by Buling-Knight almost 2 years ago
- 6 comments
#51 - 请问,mobilenetv2获取到的low_level_features里面都是0/nan是什么原因?
Issue -
State: closed - Opened by bendary about 2 years ago
- 1 comment
#50 - 请问,训练好的pth模型文件如何转换为onnx文件?
Issue -
State: open - Opened by Pengchuang about 2 years ago
- 1 comment
#49 - 训练时间长
Issue -
State: open - Opened by bulindun about 2 years ago
- 1 comment
#48 - 全都是中文,看起来太亲切了,必须给好评哈哈哈
Issue -
State: open - Opened by wqshmzh about 2 years ago
- 1 comment
#47 - 关于替换为我自己的数据集训练的问题
Issue -
State: open - Opened by Codingaworld about 2 years ago
- 2 comments
#46 - 测试结果是黑的
Issue -
State: open - Opened by mazhendi about 2 years ago
- 1 comment
#45 - 關於訓練測試的一些問題
Issue -
State: open - Opened by jay782244a about 2 years ago
- 11 comments
#44 - predict部分
Issue -
State: closed - Opened by qinshiran about 2 years ago
#43 - 判断当前batch_size,自适应调整学习率
Issue -
State: open - Opened by treeee-1 over 2 years ago
- 1 comment
#42 - 怎么在输出结果上加上标签?
Issue -
State: open - Opened by EkilGitHub over 2 years ago
- 1 comment
#41 - 模型输出问题
Issue -
State: open - Opened by YuriGao over 2 years ago
- 4 comments
#40 - 训练问题
Issue -
State: open - Opened by yangmenda over 2 years ago
- 2 comments
#39 - mae指标问题
Issue -
State: open - Opened by xiaohongzaizhe over 2 years ago
- 3 comments
#38 - 如何更换其他backbone
Issue -
State: open - Opened by tjulearning over 2 years ago
- 8 comments
#37 - Val Loss 训练后面一直往上走
Issue -
State: open - Opened by YangJing524 over 2 years ago
- 1 comment
#36 - 博主大大,请问预测过程可以调整识别率吗,可以的话对应参数在哪里呢?
Issue -
State: open - Opened by yizhipipixia over 2 years ago
- 6 comments
#35 - 关于voc数据集
Issue -
State: open - Opened by duduzai2019 over 2 years ago
- 6 comments
#34 - xception voc扩展数据集 miou
Issue -
State: open - Opened by ChouCHou-y over 2 years ago
- 5 comments
#33 - 关于语义分割Camvid数据集
Issue -
State: open - Opened by duduzai2019 over 2 years ago
- 5 comments
#32 - 输出坐标
Issue -
State: closed - Opened by SSTato over 2 years ago
- 3 comments
#31 - xception 训练miou
Issue -
State: open - Opened by Leon-0919 over 2 years ago
- 9 comments
#30 - 更换主干网络
Issue -
State: open - Opened by WEIPudding over 2 years ago
- 4 comments