Comments (11)
你好,训练和测试都在这个代码仓里,你直接运行就可以得到训练和测试的结果,
from horeid.
你好,训练和测试都在这个代码仓里,你直接运行就可以得到训练和测试的结果,
谢谢,跑出了结果
base+gcn, Dataset: Duke
mAP: 0.45936994972097783
Rank: [0.57330317 0.64434389 0.67963801 ... 1. 1. 1. ]
当我尝试接入视频提取特征,需要一张一张提取人体特征,故修改了加载数据的batchsize,运行到这里会出错
bned_feature_vector_list, cls_score_list = self.bnclassifiers(feature_vector_list)
提示:ValueError: expected 2D or 3D input (got 1D input)
修改了数据输入的batchsize
默认:
self.duke_query_loader = self._get_loader(self.duke_query_samples, self.transform_test, 128)
self.duke_gallery_loader = self._get_loader(self.duke_gallery_samples, self.transform_test, 128)
-----------------------------------
修改后:
self.duke_query_loader = self._get_loader(self.duke_query_samples, self.transform_test, 1)
self.duke_gallery_loader = self._get_loader(self.duke_gallery_samples, self.transform_test, 1)
forward的时候如果batchsize=1会报维度错误,我该如何修改,谢谢
from horeid.
我遇到了同样的问题,请问解决了吗 @nobody-cheng
from horeid.
@nobody-cheng @chenhaoCSDN
你好,请问你们在market数据集上怎么训练呢?而且在duck上的结果我也不是很准确!
from horeid.
你好,目前不支持batchsize=1, 只要bs>2应该就不会报错。
后面会修复这个bug的,谢谢。
from horeid.
@nobody-cheng @chenhaoCSDN
你好,请问你们在market数据集上怎么训练呢?而且在duck上的结果我也不是很准确!
你好,在occluded-Duke数据库上的点应该是准确的,且比论文中的点还高一些。如果结果差距很大,可能是您运行方式有错误,可以把问题反馈上来。
maket的训练代码暂时还没有整理,抱歉。
from horeid.
@wangguanan
感谢解答!是的,在occluded-Duke上的结果是会高些,我发现在duck-mtmc下的结果,反而base下的结果比base+gcn+gm还要高,请问为啥呢
from horeid.
@wangguanan
感谢解答!是的,在occluded-Duke上的结果是会高些,我发现在duck-mtmc下的结果,反而base下的结果比base+gcn+gm还要高,请问为啥呢
duke、market 的训练参数和 occluded-duke 不一样。
另外因为 duke 中 遮挡 case 很少(相比 occluded duke),因此长点不明显。
from horeid.
@wangguanan
感谢解答!是的,在occluded-Duke上的结果是会高些,我发现在duck-mtmc下的结果,反而base下的结果比base+gcn+gm还要高,请问为啥呢duke、market 的训练参数和 occluded-duke 不一样。
另外因为 duke 中 遮挡 case 很少(相比 occluded duke),因此长点不明显。
也就是说现在的仓库里只能正常得到occluded的结果了?期待作者把其他的数据集的训练参数整理下!谢啦!
from horeid.
@wangguanan
感谢解答!是的,在occluded-Duke上的结果是会高些,我发现在duck-mtmc下的结果,反而base下的结果比base+gcn+gm还要高,请问为啥呢duke、market 的训练参数和 occluded-duke 不一样。
另外因为 duke 中 遮挡 case 很少(相比 occluded duke),因此长点不明显。也就是说现在的仓库里只能正常得到occluded的结果了?期待作者把其他的数据集的训练参数整理下!谢啦!
是的,谢谢理解哈。
from horeid.
我遇到了同样的问题,请问解决了吗 @nobody-cheng
你好,请问你是在训练的时候遇到的问题,还是测试阶段。
如果是训练阶段,必须保证 bs > 1,因为 bn layer 要求 bs >1 才能计算 mean and var。
如果是测试阶段,请确认是否调用 base.set_val(),把模型设置为推理模式,这个时候 bn 会使用 running_mean and running_var, 不要求 bs >1。
如果已经调用了 base.set_val() 还报错的话,请多提供一些信息,我去看看。
抱歉最近太忙了,回复的太慢了。
from horeid.
Related Issues (20)
- implementation details of model_gcn HOT 1
- How to train on Market-1501 dataset?
- PreTrain model HOT 3
- Why not provide the code for training Market and testing on PartialReID and Partial iLIDS HOT 2
- Details about experiments on occluded datasets and partial datasets HOT 2
- Do you joint train your feature extraction backbone, HRNet and your re-id net? HOT 1
- train HOT 3
- can you share the Occulded Reid datasets link? HOT 1
- Test the datasets DukeMTMC-reID,it takes error,do you know why? HOT 1
- ValueError: expected 2D or 3D input (got 1D input) HOT 1
- 关于partial dataset以及occluded-ReID数据集的测试代码 HOT 1
- How can i train on the datasets Occluded-reid,partial-reid,Market-1501 in your paper? HOT 1
- global feature and local feature
- 请问如何使用自己的数据集进行推理?
- 请问如何使用自己的数据集进行推理?
- 您好,我用market训练的模型测试PartialREID时得到的RANK如下所示,与论文中的结果相差较大,请问测试代码是否和occluded-duke相同
- 请问在HOReID中,训练Market1501并去测试occluded-REID 参数如何设置呢?
- RuntimeError: module must have its parameters and buffers on device cuda:0 (device_ids[0]) but found one of them on device: cuda:1 HOT 2
- how to query one image when visualize mode? I try to change batch-size but occur a error HOT 3
- Padding in Transformer HOT 1
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 horeid.