PaddleDetection

PaddleDetection copied to clipboard

scores = outputs['bbox'][:, 1].numpy() RuntimeError: (PreconditionNotMet) Tensor's dimension is out of bound.Tensor's dimension must be equal or less than the size of its memory.But received Tensor's dimension is d4 memory's size is 0. [Hint: Expected numel() * SizeOfType(type()) <= memory_size(), but received numel() * SizeOfType(type()):4 > memory_size():0.] (at ..\paddle\fluid\framework\tensor.cc:39)

2

### 问题确认 Search before asking - [X] 我已经搜索过问题，但是没有找到解答。I have searched the question and found no related answer. ### 请提出你的问题 Please ask your question scores = outputs['bbox'][:, 1].numpy() RuntimeError: (PreconditionNotMet) Tensor's...

3244we

推理模型时 无法分配请求的GPU内存量

1

### 问题确认 Search before asking - [X] 我已经搜索过问题，但是没有找到解答。I have searched the question and found no related answer. ### 请提出你的问题 Please ask your question 版本：CUDA11.2/cuDNN8.2/TensorRT8.0 使用的对应版本的 inference 库 在使用TesnorRT推理模型时，出现了下列错误 我已经通过 EnableUseGpu(1000,0)，...

Shrinco

RuntimeError: ffmpeg process video: test.mp4 error

2

### 问题确认 Search before asking - [X] 我已经搜索过问题，但是没有找到解答。I have searched the question and found no related answer. ### 请提出你的问题 Please ask your question [07/26 19:18:47] ppdet.utils.checkpoint INFO: Finish resuming model...

zhuangqirong

Release/2.1

1

您好我对PR不太熟悉这里我就把我仓库修改的链接给您一下吧。 我这里主要是看到ppyolo是用paddledetection进行迭代的，然后我的项目是要对ppyolov2做了一个修改，我把ppyolov2的backbone更换为了botnet，然后原本我是打算在paddledetection的backbone的存放界面写一个botnet呢，但是因为我不知道怎么进行相关配置，就比如我写了一个botnet.py我不知道怎么在网络结构配置中引用botnet最后我就把resnet.py给改成了botnet [https://github.com/violetweir/PaddleDetection](https://github.com/violetweir/PaddleDetection) 然后我在ai studio里面写了一个项目就是用paddledetection复现botnet并作为backbone应用到项目的。 [https://aistudio.baidu.com/aistudio/projectdetail/2265776?channelType=0&channel=0](https://aistudio.baidu.com/aistudio/projectdetail/2265776?channelType=0&channel=0) 我真的十分希望能教一下怎么规范化的给paddledetection添加新功能，太谢谢了。

violetweir

Release/2.2 fuse conv bn

1

For ssd mobilenet v1 quantization training by static graph mode, the pretrained weights has been converted into weights with fusing conv and bn layers by the script of fuse_conv_bn.py ....

tonywang-sh

Release/2.2

1

For ssd mobilenet v1 quantization training by static graph mode, the pretrained weights has been converted into weights with fusing conv and bn layers by the script of fuse_conv_bn.py ....

tonywang-sh

error

2

e81OcwWezU

[PaddlePaddle Hackathon] Task 72: Add PAA

5

PR types New features PR changes APIs Describe Hi, This PR add PAA (Probalistic Anchor Assignment) implementation

gbstack

ByteTracker

3

https://github.com/PaddlePaddle/PaddleDetection/blob/67c757d3d499f8c8a03e2e380d0c091daedfb6c0/tools/infer_mot.py#L93 这里预测infer_mot 的 ByteTrack 加载权重应该是 self.model 加载 weights ，你们对应一下训练的代码中加载权重的这个。 我用的配置文件是 mot/bytetrack/yoloe....xxx.yaml

Xingyu-Romantic

update ppyoloe export exclude_nms

1

sum -> concat

yghstill