RenJianZuiDeYi

模型转换的问题

2

### Before Asking - [X] I have read the [README](https://github.com/meituan/YOLOv6/blob/main/README.md) carefully. 我已经仔细阅读了README上的操作指引。 - [X] I want to train my custom dataset, and I have read the [tutorials for training your...

**Describe the bug** 在eqlv2_loss.py中，121行代码感觉是有语法问题： `def forward(self, cls_score: Tensor, label: Tensor, weight: Optional[Tensor] = None, avg_factor: Optional[int] = None, reduction_override: Optional[Tensor] = None) -> Tensor: """`Equalization Loss v2 `_ Args: cls_score...

Adopting different image sampling strategies

1

### Search before asking - [X] I have searched the YOLOv5 [issues](https://github.com/ultralytics/yolov5/issues) and [discussions](https://github.com/ultralytics/yolov5/discussions) and found no similar questions. ### Question If I want to sample images using different sampling...

Improved yolov6 to have the same trt model inference speed under int8 and fp16

2

## Description I improved yolov6. After converting to tensorrt, the improved model's inference speed is faster than the original yolov6 under fp32 and fp16. However, after converting to int8, the...

Thank you for your good job!I trained on the nuscenes-mini dataset for 20 epochs, but the output of each epoch is only about 0.0001. Is there something wrong with my...

pt转trt后为什么速度提升不大？

1

pt模型下，在YOLOv5中加了rcsosa模块后比原始yolov5精度高、速度快； 但转成tensorrt后，速度比原始yolov5慢很多，请问是什么原因？

请问RCS-OSA模块替换YOLOv5中的C3模块是否可以提升检测速度？

2

首先非常感谢您的工作！在您的论文中提到在RCS-YOLO、RepVGG-CSP的消融实验，想请问一下是否是在相同的位置采用BottleneckCSPC、RCS-OSA两个模块进行实验对比？如果是我认为的这样，BottleneckCSPC仅比C3模块多一个conv1x1，那在YOLOv5中，采用RCS-OSA替换C3模块应该也可以提升模型的检测性能。请问您是否做过RCS-OSA有效性的验证实验？该模块是否能提升检测速度？

map is too low.

1

Hi,I trained on the nuscenes-mini dataset for 200 epochs, and only modified the dataset path on the default configuration file, but the accuracy is still only 0.0001, which is the...

怎样采用Pconv+Pwconv替换常规的3x3卷积？

4

我在YOLOv5中采用Pconv（输入通道是c）+conv1x1（输入通道c,输出通道2c）组合的方式替换conv3x3（输入通道是c、输出通道是2c），特征图尺度是没有改变的，测试后精度和速度均降低了；我想请问是需要采用Fasternet Block进行来替换conv3x3嘛？它的输入通道、输出通道该怎么设置呢？是否能够提升检测速度？