Comments (4)
参考的:https://github.com/Try2ChangeX/NanoTrack_RK3588_python 项目利用这个项目的模型文件,直接用python跑是正常的,然后参考您之前那个lighttrack的项目和nanotrack的ncnn项目做的,目前的问题是跟踪结果完全不正确,下附后处理部分的代码链接:https://github.com/fly-miao/issure_code
可能存在的问题:
1.模型是未量化的,是fp16
backbone_T.rknn的参数:
model input num: 1, output num: 1
index=0, name=x.3, n_dims=4, dims=[1, 127, 127, 3], n_elems=48387, size=96774, fmt=NHWC, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
index=0, name=975, n_dims=4, dims=[1, 48, 8, 8], n_elems=3072, size=6144, fmt=NCHW, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
model is NHWC input fmt
model is NCHW output fmt
backbone_X.rknn的参数:
model input num: 1, output num: 1
index=0, name=x.3, n_dims=4, dims=[1, 255, 255, 3], n_elems=195075, size=390150, fmt=NHWC, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
index=0, name=975, n_dims=4, dims=[1, 48, 16, 16], n_elems=12288, size=24576, fmt=NCHW, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
model is NHWC input fmt
model is NCHW output fmt
head.rknn的参数:
model input num: 2, output num: 2
index=0, name=z_f.1, n_dims=4, dims=[1, 8, 8, 48], n_elems=3072, size=6144, fmt=NHWC, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
index=1, name=x_f.1, n_dims=4, dims=[1, 16, 16, 48], n_elems=12288, size=24576, fmt=NHWC, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
index=0, name=231, n_dims=4, dims=[1, 2, 16, 16], n_elems=512, size=1024, fmt=NCHW, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
index=1, name=26, n_dims=4, dims=[1, 4, 16, 16], n_elems=1024, size=2048, fmt=NCHW, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
model is NHWC input fmt
model is NCHW output fmt
2.我看这个的python项目中没有bgr转rgb的操作,我默认使用bgr输入(也试过rgb,结果也是不对)
3.后处理部分,我完全按照lighttrack那个项目的操作没有改
from lighttrack-ncnn.
有没有对比python版本推理的output与C++版本的output是否有一样?
from lighttrack-ncnn.
I'm having the same problems, is there any progress?
from lighttrack-ncnn.
When I examine the models one by one in Netron App, I see that the inputs of the models are in NCHW format. However, when I load the model using the RKNPU2 C++ APIs, I see that the inputs are in NHWC format. Therefore, I believe that the implementation is not working correctly. How can I resolve this issue?
from lighttrack-ncnn.
Related Issues (14)
- 编译和运行的问题 HOT 10
- 怎么用gpu进行推理? HOT 3
- 是否进行过量化 HOT 6
- 训练自己的数据集 HOT 1
- 按照教程转pth为onnx格式,报错Process finished with exit code 132 (interrupted by signal 4: SIGILL) HOT 9
- 你好大佬,我打开跑这个模型在rk3588开发板上。 HOT 8
- 编译报错 HOT 2
- 太厉害了吧也
- 请问 LightTrack-rknn工程中onnx转rknn模型的时候,你的3个模型是如何归一化的 HOT 1
- ncnn HOT 11
- LightTrack在rk3588上部署 HOT 4
- run ./LightTrack报错 HOT 2
- pscore_window max score一直为nan HOT 5
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
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.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lighttrack-ncnn.