update README for superpoint

Author: xmba15

README.md

@@ -8,7 +8,7 @@ Hope that they both are helpful for your work.
 ## TODO
 
 - [x] Support inference of multi-inputs, multi-outputs
-- [x] Examples for famous models, like yolov3, mask-rcnn, [ultra-light-weight face detector](https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB), [yolox](https://github.com/Megvii-BaseDetection/YOLOX), [PaddleSeg](https://github.com/PaddlePaddle/PaddleSeg/tree/release/2.3). Might consider supporting more if requested
+- [x] Examples for famous models, like yolov3, mask-rcnn, [ultra-light-weight face detector](https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB), [yolox](https://github.com/Megvii-BaseDetection/YOLOX), [PaddleSeg](https://github.com/PaddlePaddle/PaddleSeg/tree/release/2.3), [SuperPoint](https://github.com/magicleap/SuperPointPretrainedNetwork). Might consider supporting more if requested
 - [ ] Batch-inference
 
 ## Installation
@@ -228,3 +228,35 @@ wget https://github.com/Megvii-BaseDetection/YOLOX/releases/download/0.1.1rc0/yo
     +  test result on a new scene at Odaiba, Tokyo, Japan
 
 ![paddleseg odaiba](./data/images/odaiba_result.jpg)
+
+### [SuperPoint](https://arxiv.org/pdf/1712.07629.pdf)
+
+---
+
+![super_point_good_matches](./data/images/super_point_good_matches.jpg)
+
+<details>
+
+- Convert SuperPoint's pretrained weights to onnx format
+
+```bash
+git submodule update --init --recursive
+python3 -m pip install -r scripts/superpoint/requirements.txt
+python3 scripts/superpoint/convert_to_onnx.py
+```
+
+- Download test images from [this dataset](https://github.com/StaRainJ/Multi-modality-image-matching-database-metrics-methods)
+
+```bash
+wget https://raw.githubusercontent.com/StaRainJ/Multi-modality-image-matching-database-metrics-methods/master/Multimodal_Image_Matching_Datasets/ComputerVision/CrossSeason/VisionCS_0a.png -P data
+
+wget https://raw.githubusercontent.com/StaRainJ/Multi-modality-image-matching-database-metrics-methods/master/Multimodal_Image_Matching_Datasets/ComputerVision/CrossSeason/VisionCS_0b.png -P data
+```
+
+- Test inference apps
+
+```bash
+./build/examples/super_point ./scripts/superpoint/super_point.onnx data/VisionCS_0a.png data/VisionCS_0b.png
+```
+
+</details>

data/images/super_point_good_matches.jpg

@@ -86,6 +86,8 @@ int main(int argc, char* argv[])
                     cv::Scalar::all(-1), cv::Scalar::all(-1), std::vector<char>(),
                     cv::DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS);
     cv::imwrite("super_point_good_matches.jpg", matchesImage);
+    cv::imshow("super_point_good_matches", matchesImage);
+    cv::waitKey();
 
     return EXIT_SUCCESS;
 }
@@ -199,7 +201,6 @@ KeyPointAndDesc processOneFrame(const Ort::SuperPoint& osh, const cv::Mat& input
                                 float confidenceThresh, bool alignCorners, int distThresh)
 {
     int origW = inputImg.cols, origH = inputImg.rows;
-    std::vector<float> originImageSize{static_cast<float>(origH), static_cast<float>(origW)};
     cv::Mat scaledImg;
     cv::resize(inputImg, scaledImg, cv::Size(Ort::SuperPoint::IMG_W, Ort::SuperPoint::IMG_H), 0, 0, cv::INTER_CUBIC);
     osh.preprocess(dst, scaledImg.data, Ort::SuperPoint::IMG_W, Ort::SuperPoint::IMG_H, Ort::SuperPoint::IMG_CHANNEL);

examples/SuperPointApp.cpp

@@ -12,9 +12,8 @@
 #pragma once
 
 #include <algorithm>
-#include <opencv2/core/fast_math.hpp>
-#include <stdexcept>
 #include <string>
+#include <utility>
 #include <vector>
 
 #include <opencv2/opencv.hpp>

examples/Utility.hpp

@@ -37,6 +37,7 @@ def main():
             "output": {0: "batch_size"},
         },
     )
+    print(f"\nonnx model is saved to: {os.getcwd()}/super_point.onnx")
 
 
 if __name__ == "__main__":