Kaeseknacker
Kaeseknacker
**Describe the bug** I installed the newest MMDetection version but can not convert a simple frcnn model. ``` mim download mmdet --config faster_rcnn_x101_64x4d_fpn_1x_coco --dest . mmdet2trt --save-engine=true --min-scale 1 3...
**Describe the bug** I have a problem with a (fp16) converted model which produces wrong detection results. In some boxes, the x value in the top left corner is 0,...
Hi, I'm trying to train a SR-ResNet for scale = 2, but unfortunately this always fails. As I see it, the SR-ResNet implementation only works for scaling factor = 4....
The NMS in the preprocessing step produces different output from the original implementation in python: https://github.com/apennisi/deep_sort/blob/2fefac0734197c99114055f4257668869b4f0834/src/features/src/model.cpp#L104 NMS can be replaced with this implementation: https://github.com/martinkersner/non-maximum-suppression-cpp Maybe it is also faster.
`_cosine_distance` implementation is incomplete (normalization is missing). Can be fixed with these lines of code: ```cpp Eigen::MatrixXf NearNeighborDisMetric::_cosine_distance(const FEATURESS& x, const FEATURESS& y, bool data_is_normalized) { FEATURESS a = x;...
I think the euclidian distance is calculated wrong: https://github.com/apennisi/deep_sort/blob/2fefac0734197c99114055f4257668869b4f0834/src/matching/src/nn_matching.cpp#L138 ```cpp Eigen::VectorXf NearNeighborDisMetric::_nneuclidean_distance( const FEATURESS &x, const FEATURESS &y) { MatrixXf distances = _pdist(x,y); VectorXf res = distances.colwise().maxCoeff().transpose(); res = res.array().max(VectorXf::Zero(res.rows()).array());...
### Checklist - [X] I have searched related issues but cannot get the expected help. - [X] 2. I have read the [FAQ documentation](https://github.com/open-mmlab/mmdeploy/tree/main/docs/en/faq.md) but cannot get the expected help....