-
### 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
### 使用8卡训练 ppyoloe s 300个epoch,日志中显示训练期间从未显示eva…
-
### 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
![image](https://user-images.githubuserconten…
-
### 问题确认 Search before asking
- [X] 我已经查询[历史issue](https://github.com/PaddlePaddle/PaddleDetection/issues),没有发现相似的bug。I have searched the [issues](https://github.com/PaddlePaddle/PaddleDetection/issu…
-
训练时分pink、black、blue,3个种类的标签,最终训练结果map接近100%。
部署过后predict的结果不管用什么照片,只有black一种标签
![image](https://github.com/user-attachments/assets/2b36d317-7bca-4b9b-946a-ae0cbf31cf71)
![image](https://github.com/…
-
## 环境
- 【FastDeploy版本】: fastdeploy- Windows-gpu-1.0.6
- 【系统平台】: Windows x64(Windows10)
- 【硬件】: 笔记本电脑 Nvidia GPU 2060 6G显存, CUDA 11.2 CUDNN 8.2
- 【编译语言】: Python(3.9)
## 问题
- 【性能问题】
- - 在使…
-
在python3.8环境中,用下面的代码把ppyoloe转化固定shape的onnx模型量化转为rknn报错:
# Build model
print('--> Building model')
ret = rknn.build(do_quantization=True, dataset=DATASET) #量化,会报错
# ret = rknn.build(do_quantization…
-
### 问题确认 Search before asking
- [X] 我已经查询[历史issue](https://github.com/PaddlePaddle/PaddleDetection/issues),没有发现相似的bug。I have searched the [issues](https://github.com/PaddlePaddle/PaddleDetection/issu…
-
### 问题确认 Search before asking
- [X] 我已经查询[历史issue](https://github.com/PaddlePaddle/PaddleDetection/issues),没有发现相似的bug。I have searched the [issues](https://github.com/PaddlePaddle/PaddleDetection/issu…
-
**PaddleDetection team appreciate any suggestion or problem you delivered~**
## Checklist:
1. 查找历史相关issue寻求解答/I have searched related issues but cannot get the expected help.
2. 翻阅[FAQ](https:/…
-
### 问题描述 Please describe your issue
bytetrack 支持c++部署吗