tracker/nn: Tweaks, refactoring, a deadzone filtering and support for uncertainty estimation

* Add rudimentary test for two functions .. maybe more in future
* Fix the rotation correction from vertical translation
* Move preview class to new files
* Move neural network model adapters to new files
* Add utility functions for opencv
* Query the model inputs/outputs by name to see what is available
* Supports outputs for standard deviation of the data distribution -
  What you get if you let your model output the full parameters of a
  gaussian distribution (depending on the inputs) and fit it with
  negative log likelihood loss.
* Disabled support for sequence models
* Add support for detection of eye open/close classification.
  Scale uncertainty estimate up if eyes closed
* Add a deadzone filter which activates if the model supports uncertainty
  quantification. The deadzone scales becomes larger the more uncertain
  the model/data are. This is mostly supposed to be useful to suppress
  large estimate errors when the user blinks with the eyes
* Fix distance being twice of what it should have been

1 files changed, 60 insertions, 0 deletions

diff --git a/tracker-neuralnet/preview.h b/tracker-neuralnet/preview.h
new file mode 100644
index 00000000..adc12993
--- /dev/null
+++ b/tracker-neuralnet/preview.h
@@ -0,0 +1,60 @@
+/* Copyright (c) 2021 Michael Welter <michael@welter-4d.de>
+ *
+ * Permission to use, copy, modify, and/or distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ */
+
+#pragma once
+
+#include "model_adapters.h"
+
+#include "cv/video-widget.hpp"
+
+#include <optional>
+
+#include <opencv2/core.hpp>
+#include <opencv2/imgproc.hpp>
+
+
+namespace neuralnet_tracker_ns
+{
+
+/** Makes a maximum size cropping rect with the given aspect. 
+*   @param aspect_w: nominator of the aspect ratio
+*   @param aspect_h: denom of the aspect ratio
+*/
+cv::Rect make_crop_rect_for_aspect(const cv::Size &size, int aspect_w, int aspect_h);
+
+
+/** This class is responsible for drawing the debug/info gizmos
+* 
+* In addition there function to transform the inputs to the size of
+* the preview image which can be different from the camera frame.
+*/
+class Preview
+{
+public:
+    void init(const cv_video_widget& widget);
+    void copy_video_frame(const cv::Mat& frame);
+    void draw_gizmos(
+        const std::optional<PoseEstimator::Face> &face,
+        const std::optional<cv::Rect2f>& last_roi,
+        const std::optional<cv::Rect2f>& last_localizer_roi,
+        const cv::Point2f& neckjoint_position);
+    void overlay_netinput(const cv::Mat& netinput);
+    void draw_fps(double fps, double last_inference_time);
+    void copy_to_widget(cv_video_widget& widget);
+private:
+    // Transform from camera image to preview
+    cv::Rect2f transform(const cv::Rect2f& r) const;
+    cv::Point2f transform(const cv::Point2f& p) const;
+    float transform(float s) const;
+
+    cv::Mat preview_image_;
+    cv::Size preview_size_ = { 0, 0 };
+    float scale_ = 1.f;  
+    cv::Point2f offset_ = { 0.f, 0.f};
+};
+
+} // neuralnet_tracker_ns
\ No newline at end of file