Mikhail Scherbina

MMPose segfault at the last feature

1

https://github.com/linghu8812/tensorrt_inference/blob/master/code/src/mmpose.cpp#L75 ```c++ float *current_person = output + index * out_size; for (int number = 0; number < m_num_key_points; number++) { float *current_point = current_person + feature_size * number; auto max_pos...

Memory leak

1

[This code contains](https://github.com/WongKinYiu/yolov7/blob/pose/models/yolo.py#L377) a memory leak. I tried looking at memory usage with nvidia-smi, and with each loop pass this function consumes more and more memory without it being free'd....

disable npp in multistream context

10

### Pull Request Readiness Checklist See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request - [* ] I agree to contribute to the project under Apache 2 License. - [* ] To the best of...

NPP performance bottleneck with multiple streams

5

##### System information (version) - OpenCV => 4.6 - Operating System / Platform => Linux x64 with Nvidia Cuda 11.6 - Compiler => gcc ##### Detailed description Hello! I am...

Some types are disabled by default due to being incompatible with Mypy can't go away

1

## Bug description Here is my prisma config: ``` generator client { provider = "prisma-client-js" } generator py { provider = "prisma-client-py" interface = "sync" recursive_type_depth = -1 } ```...

Lower n

3

**How should the input text representation look like** > Example: y_n **How should the ASCII art output look like** > Example: y lower n, you can imagine **Describe in what...

Back Surfaces Culling Mistake

2

Hello! I believe, back surfaces culling should be done in screen coordinates! Not in the original coordinates. Let's discuss the implications doing original coordinates culling may cause. For example, I...

add data in matches for doppelganger

2

This proposal will allow for seamless integration with the [Doppelgangers](https://github.com/RuojinCai/doppelgangers) repository.

Doppelgangers support

5

There is a repository, which aims to avoid incorrect image registration called 'Doppelgangers': [link](https://github.com/RuojinCai/doppelgangers). Implementing it into the pipeline should be straightforward, since it is a binary classifier, also I...

Well, the title speaks for itself.