Git Product home page Git Product logo

nnie-lite's Issues

ERROR: "caffe.LayerParameter" has no field named "bn_param"

I train Enet with the repo https://github.com/TimoSaemann/ENet, finnally I got prototxt and model weights, but it has
bn_param { scale_filler { type: "constant" value: 1.0 } shift_filler { type: "constant" value: 0.0010000000474974513 } bn_mode: INFERENCE }
in the bn_conv_merged_model.prototxt, it will meets error,
./nnie_mapper_12 model_inst.cfg Mapper Version 1.2.2.0_B010 (NNIE_1.2) 19050917062183 begin net parsing.... [libprotobuf ERROR google/protobuf/text_format.cc:307] Error parsing text-format caffe.NetParameter: 65:12: Message type "caffe.LayerParameter" has no field named "bn_param".
How do you deal with bn_param?

关于yolov5的移植问题

我在作者的基础上修改一些代码,移植好了yolov5,但目前有些问题没搞明白:
用high-precision量化的wk跑起来仍然有一些精度损失,主要表现为confidence略有下降,对于本身得分就不高的目标容易发生漏检,请问这个现象是合理吗?还是说哪里弄得有问题,现在我也在检查很久了,没发现问题。

ps:(最好的办法是每一层打印输出和torch模型对比,但可能我太菜了。。觉得这个工作量有点难受)

yolov5s部署到hi3521dv200出错

你好,我把你的代码进行了部分更改,部署到hi3521dv200上,在进行模型推理的时候报错:

[Level]:Error,[Func]:NNIE_Forward [Line]:203 [Info]:Error,HI_MPI_SVP_NNIE_Forward failed!
[Level]:Error,[Func]:NNIE_Forward_From_Data [Line]:331 [Info]:Error,SAMPLE_SVP_NNIE_Forward failed!

能请教下是哪方面出现问题了吗

ENet的demo编译后执行有报错,看起来是模型的关系,是否有解决的思路?

报错信息如下

[Func]:HI_MPI_SYS_MmzAlloc_Cached [Line]:1044 [Info]:System alloc mmz memory failed!
[Level]:Error,[Func]:SAMPLE_SVP_NNIE_ParamInit [Line]:355 [Info]:Error,Malloc memory failed!

[Level]:Error,[Func]:SAMPLE_COMM_SVP_NNIE_ParamInit [Line]:466 [Info]:Error, SAMPLE_SVP_NNIE_ParamInit failed!

[Level]:Debug,[Func]:SAMPLE_COMM_SVP_CheckSysExit [Line]:95 [Info]:Svp mpi exit ok!
[Level]:Info,[Func]:nnie_param_init [Line]:34 [Info]:Error,nnie_Param_init failed!
Segmentation fault

数据输入,量化的问题

1.假设我从摄像头获取数据,VPSS可以直接输出resize后的YUV数据,理论可以省掉opencv的数据处理层。
2.输入端为8bit 正整数,一般的模型均为float32,这样的条件下,如果做量化效果会比较好

detect 检测时显示初始化失败

[Level]:Error,[Func]:SAMPLE_COMM_SVP_SysInit [Line]:42 [Info]:Error(0xa0018012):HI_MPI_VB_SetConf failed!

[Level]:Error,[Func]:SAMPLE_COMM_SVP_CheckSysInit [Line]:76 [Info]:Svp mpi init failed!

ParseYolov3Feature计算框和类别的得分问题

你好!想问下utils.h里ParseYolov3Feature函数里面:
108行float tc = feature.data[cx + (cy * feature.width) + (channel + 4) * feature.height * feature.width];应该是目标框的得分,然后
123行用Softmax计算每个类别的得分,但是yolov3之后不是用Sigmoid计算每个框的得分吗,这样可以实现多标签分类,同一个框可以说不同类别的物体。我看海思给的yolov3 demo里,在sample_svp_nnie_software.c,SVP_NNIE_Yolov3_GetResult函数里也是用Sigmoid计算框和各类别的得分的

NNIE_Get_Result 偶尔会输出远超4096的值

网络多数时候运行是正常的,偶尔会输出远超4096的值。
https://github.com/RaySue/NNIE-lite/blob/daf0dc19f47bc1e286308d137e1e636c2ef98da7/src/nnie_core.c#L272

                for (n = 0; n < pstNnieParam->astSegData[u32SegIdx].astDst[u32NodeIdx].u32Num; n++)
                {
                    for (i = 0; i < u32Chn; i++)
                    {
                        for (j = 0; j < u32Height; j++)
                        {
                            for (k = 0; k < u32Width; k++)
                            {
                                *(tmp + k) = (*(ps32ResultAddr + k)) * 1.0f / SAMPLE_SVP_NNIE_QUANT_BASE;
                            }
                            ps32ResultAddr += u32Stride / sizeof(HI_U32);
                            tmp += t.width;
                        }
                    }
                }

对*(ps32ResultAddr + k)打印,偶尔会出现特别大的值,远超4096

能给出yolov4后处理部分的实现代码吗?

感谢您的工作和奉献,这个工程您还在继续维护吗?能给出yolov4后处理部分代码吗?我调试了很久,yolov4的检测效果很差,觉得是和后处理部分代码有关。

yolov4推理不准确

大佬你好,我在海思上转换了一个yolov4的模型,利用NNIE-LITE进行推理,测试发现识别不准确,如图:
image

我的anchors配置:
image
其他代码我未做修改,这是为什么呢?
NNIE转换钱的caffe模型在PC上测试检测准确,盼望大佬解决。

执行编译好的文件会报错!

您好,按照您的配置编译好的二进制文件seg_exe,在terminal执行./seg_exe会报错
'
bash: ./alg_Exe:无法执行二进制文件: 可执行文件格式错误
'
请问该如何解决呢?非常感谢

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.