Yang Nie

add MobileNeXt

1

1. add MobileNeXt 2. fix typo in test_tipc/docs/install.md 3. fix DecodeImage pil backend 4. change the default interpolation of pil from NEAREST to BILINEAR (same to cv2 backend)

### Prerequisite - [X] I have searched [Issues](https://github.com/open-mmlab/mmrotate/issues) and [Discussions](https://github.com/open-mmlab/mmrotate/discussions) but cannot get the expected help. - [X] I have read the [FAQ documentation](https://mmrotate.readthedocs.io/en/1.x/notes/4_faq.html) but cannot get the expected help....

[Feature] Support RT-DETR

12

## Motivation Support [RT-DETR](https://arxiv.org/abs/2304.08069) as discussed in [this issue](https://github.com/open-mmlab/mmdetection/issues/10186) Referred to the following repositories for implementation details: - [lyuwenyu/RT-DETR](https://github.com/lyuwenyu/RT-DETR) - [PaddlePaddle/PaddleYOLO](https://github.com/PaddlePaddle/PaddleYOLO/tree/develop/configs/rtdetr) - [PaddlePaddle/PaddleDetection](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.7/configs/rtdetr) - [nijkah/mmdetection](https://github.com/nijkah/mmdetection/tree/rtdetr) in [PR #10498](https://github.com/open-mmlab/mmdetection/pull/10498) ## Modification...

Fix wrong size of mask in RTMDet-Ins post-processing.

1

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand...

refactor RandAugment, RandAugmentV2 and RandAugmentV3

2

1. 重构前后`RandAugment`的行为有一些不同：重构前`translateX`,`translateY`和`rotate`未指定`resample`方式（默认为`NEAREST`），重构后指定`resample`为`BICUBIC`。 2. 原先的`RandAugmentV2`中，`progress_magnitude`参数未起作用，这次重构暂时注释掉了。

fp16 overflow in CrossEntropyLossCost

1

Thanks for your error report and we appreciate it a lot. **Checklist** 1. I have searched related issues but cannot get the expected help. 2. I have read the [FAQ...

### 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/issues) and found no similar bug report. ### Bug组件 Bug Component Training ### Bug描述 Describe the Bug https://github.com/PaddlePaddle/PaddleDetection/blob/dbb9f5b2ccd99a68521527d63dcc3cc4bffc94c9/ppdet/engine/trainer.py#L552 在training时，有一些数据增强需要epoch来决定是否生效...

无法复现精度

5

你好，我使用您的代码在coco上进行复现，无法得到你所标称的精度，且差距较大。 我只在代码中修改coco的路径，其他部分未修改。 使用配置`configs/dfine/dfine_hgnetv2_l_coco.yml`训练，得到的精度为`53.5`

## 问题描述 当前PaddleClas中，多卡训练时seed未被正确设置，会导致dataloader的不同线程使用相同的random seed。 在ppcls/engine/engine.py中： ``` if seed or seed == 0: paddle.seed(seed) np.random.seed(seed) random.seed(seed) ... # build dataloader here self.train_dataloader = build_dataloader(self.config["DataLoader"], "Train", self.device, self.use_dali) ... if self.config["Global"]["distributed"]: ......