zihaomu

Bug fixed of s1d1 in NN Conv

2

Fix the critical error in `s1d1`. It can pass all the regression test in `opencv_extra`. And I will update the `s1d1` of FP16 tomorrow.

DNN: Reduce Layer (add dynamic batch and ReduceSum support)

1

Related ReduceSum issue #22195 Related Dynamic Batch of Reduce Layer issue #22086 In this PR, we supported two input of `ReduceSum layer` and dynamic batch size in `Reduce Layer` of...

DNN: load fp16 ONNX model as fp32

2

The FP16 model is half the size of the FP32 model, which will be edge device friendly. Currently, OpenCV `ONNX_importer` cannot load FP16 models correctly and DNN also doesn't have...

DNN: Add qgemm and squeeze op13 supported on ONNXImporter

1

[Squeeze node of ONNX](https://github.com/onnx/onnx/blob/main/docs/Operators.md#squeeze) has moved the axes from attribute to input. [Test Data](https://github.com/opencv/opencv_extra/pull/989) ### Pull Request Readiness Checklist See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request - [x] I agree to contribute to...

DNN: FP16 support on Convolution 2D - WIP

5

## FP16 support on ARM platform This PR proposes to support FP16 backend in Convolution. For now, we only support FP16 at ARM aarch64. ### TODO List | Task |...

Hi teams, The YoloV4 model at [here](https://github.com/onnx/models/tree/main/vision/object_detection_segmentation/yolov4) is too complicated and hard to be optimized. And I have generated a new [YoloV4 model](https://drive.google.com/file/d/1HRv5lrnd0qJIMxdEnE118TMlSh3Xj2Ht/view?usp=sharing) by `onnx-simplifier` based on your Yolov4. Is...

Related PR [#22337](https://github.com/opencv/opencv/pull/22337).

[related PR](https://github.com/opencv/opencv/pull/21415)

Fix possible out-of-bounds bug in OpConv.fx

2

Fix possible out-of-bounds bug