Query on Cluster Visualization about depth_clustering

[niosus]

@vin-cloud the whole visualization pipeline could see some improvements, I would write it better now 😉 But now there is no more time for that. What you are looking for lives here

depth_clustering/src/qt/drawables/object_painter.cpp

Lines 77 to 101 in fe9d667

	Drawable::UniquePtr ObjectPainter::CreateDrawablePolygon3d(
	const depth_clustering::Cloud& cloud) {
	float min_z{std::numeric_limits<float>::max()};
	float max_z{std::numeric_limits<float>::lowest()};
	std::vector<cv::Point2f> cv_points;
	cv_points.reserve(cloud.size());
	std::vector<int> hull_indices;
	for (const auto& point : cloud.points()) {
	cv_points.emplace_back(cv::Point2f{point.x(), point.y()});
	min_z = std::min(min_z, point.z());
	max_z = std::max(max_z, point.z());
	}
	cv::convexHull(cv_points, hull_indices);
	DrawablePolygon3d::AlignedEigenVectors hull;
	hull.reserve(cloud.size());
	for (int index : hull_indices) {
	const auto& cv_point = cv_points[index];
	hull.emplace_back(cv_point.x, cv_point.y, min_z);
	}
	const float diff_z = max_z - min_z;
	if (diff_z < 0.3) {
	return nullptr;
	}
	return DrawablePolygon3d::Create(hull, diff_z);
	}

I just quickly implementer an OpenCV-based convex hull computation to improve the visualization.

[niosus]

@vin-cloud I will close this issue for now. If you'd rather leave it open, feel free to reopen again.