CorNet: Generic 3D Corners for 6D Pose Estimation of New Objects without Retraining

Abstract

We present a novel approach to the detection and 3D pose estimation of objects in color images. Its main contribution is that it does not require any training phases nor data for new objects, while state-of-the-art methods typically require hours of training time and hundreds of training registered images. Instead, our method relies only on the objects' geometries. Our method focuses on objects with prominent corners, which covers a large number of industrial objects. We first learn to detect object corners of various shapes in images and also to predict their 3D poses, by using training images of a small set of objects. To detect a new object in a given image, we first identify its corners from its CAD model; we also detect the corners visible in the image and predict their 3D poses. We then introduce a RANSAC-like algorithm that robustly and efficiently detects and estimates the object's 3D pose by matching its corners on the CAD model with their detected counterparts in the image. Because we also estimate the 3D poses of the corners in the image, detecting only 1 or 2 corners is sufficient to estimate the pose of the object, which makes the approach robust to occlusions. We finally rely on a final check that exploits the full 3D geometry of the objects, in case multiple objects have the same corner spatial arrangement. The advantages of our approach make it particularly attractive for industrial contexts, and we demonstrate our approach on the challenging T-LESS dataset.

Toward Augmented Reality in Museums: Evaluation of Design Choices for 3D Object Pose Estimation

“…The pose is recovered in post-processing using RANSAC and PnP. CorNet (Pitteri et al, 2019) proposes an oriented corner detector for estimating the pose of industrial objects. A FasterRCNN (Ren et al, 2017) network is used as the corner detector.…”

Perspective-2-Ellipsoid: Bridging the Gap Between Object Detections and 6-DoF Camera Pose

“…This does not allow to conduct any fair comparison. We nevertheless compare our method to CorNet [11], which aims at computing pose from recognized bbox ellipses GT ellipses generic 3D corners without specific scene retraining. Results are available on object 20 from scene 08 (T-LESS).…”

“…Indeed, object detection algorithms are able to recognize objects across a wide range of viewpoints and in different weather or lighting conditions. This opens the way towards more robust pose algorithms based on highlevel features (objects or corners [11]) instead of traditional low-level primitives (keypoints). Li et al [12], [13] proposed to use object detections to estimate relative camera poses in the case of large changes in viewpoints.…”

I Like to Move It: 6D Pose Estimation as an Action Decision Process

“…If training is is done for multiple objects, the resulting predictions become unreliable [33]. The recent CorNet [58] focuses on objects geometry instead and detects object-agnostic corners. While this is more robust, it is in spirit similar to early pose estimation approaches that detect significant points.…”