Abstract: A method for 2D semantic keypoint detection and tracking is described. The method includes learning embedded descriptors of salient object keypoints detected in previous images according to a descriptor embedding space model. The method also includes predicting, using a shared image encoder backbone, salient object keypoints within a current image of a video stream. The method further includes inferring an object represented by the predicted, salient object keypoints within the current image of the video stream. The method also includes tracking the inferred object by matching embedded descriptors of the predicted, salient object keypoints representing the inferred object within the previous images of the video stream based on the descriptor embedding space model.

Abstract: A method for discovering human-interpretable concepts from video-based transformer models is described. The method includes passing a set of videos through a video-based transformer model to select an intermediate video feature of each of the set of videos. The method also includes clustering the intermediate video feature of each of the set of videos to obtain corresponding tubelets to the selected intermediate video features of each of the set of videos. The method further includes clustering an entire dataset of tubelets to form concepts of the set of videos. The method also includes calculating an importance of each of the concepts of the set of videos to an output of the video-based transformer model.

Abstract: A method for 3D object detection is described. The method includes predicting, using a trained monocular depth network, an estimated monocular input depth map of a monocular image of a video stream and an estimated depth uncertainty map associated with the estimated monocular input depth map. The method also includes feeding back a depth uncertainty regression loss associated with the estimated monocular input depth map during training of the trained monocular depth network to update the estimated monocular input depth map. The method further includes detecting 3D objects from a 3D point cloud computed from the estimated monocular input depth map based on seed positions selected from the 3D point cloud and the estimated depth uncertainty map. The method also includes selecting 3D bounding boxes of the 3D objects detected from the 3D point cloud based on the seed positions and an aggregated depth uncertainty.

Abstract: A method for estimating depth of a scene includes capturing a first image of the scene via one or more sensors associated with a first agent. The method also includes selecting one or more second images from a group of previously captured images of the scene, each second image of the one or more second images satisfying a depth criteria, each image of the group of previously captured images being captured prior to the first image. The method further includes estimating the depth of the scene based on the first image and the one or more second images.

Abstract: A method for fusing neural radiance fields (NeRFs) is described. The method includes re-rendering a first NeRF and a second NeRF at different viewpoints to form synthesized images from the first NeRF and the second NeRF. The method also includes inferring a transformation between a re-rendered first NeRF and a re-rendered second NeRF based on the synthesized images from the first NeRF and the second NeRF. The method further includes blending the re-rendered first NeRF and the re-rendered second NeRF based on the inferred transformation to fuse the first NeRF and the second NeRF.

Abstract: A method for generating a visual representation of an environment based on a point cloud includes hierarchically processing the point cloud with different granularity levels to generate multiple groups of primitives and multiple sets of points. The method also includes generating a group of intermediate sets associated with the point cloud, each intermediate set associated with one of the multiple groups of primitives and one of the multiple sets of points, having a same granularity level. The method further includes iteratively determining respective features associated with each intermediate set of a sequence of intermediate sets, each intermediate set included the set of primitives and the set of points, the respective features including first features of the set of primitives and second features of the set of points. The method still further includes generating the visual representation based on the respective features of each one of the sequence of intermediate sets.

Abstract: A method for scale-aware depth estimation using multi-camera projection loss is described. The method includes determining a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes training a scale-aware depth estimation model and an ego-motion estimation model according to the multi-camera photometric loss. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the scale-aware depth estimation model and the ego-motion estimation model. The method also includes planning a vehicle control action of the ego vehicle according to the 360° point cloud of the scene surrounding the ego vehicle.

Abstract: A method for labeling keypoints includes labeling, via a keypoint model, a first set of keypoints in a first image associated with a three-dimensional (3D) map of an environment, the first image having been captured during a first time period. The method also includes labeling, via the keypoint model, a second set of keypoints in a second image associated with the 3D map, the second image having been captured during a second time period. The method further includes updating, via the keypoint model, one of more of the first set of keypoints based on labeling the second set of keypoints.

Abstract: A method for improving 3D object detection via object-level augmentations is described. The method includes recognizing, using an image recognition model of a differentiable data generation pipeline, an object in an image of a scene. The method also includes generating, using a 3D reconstruction model, a 3D reconstruction of the scene from the image including the recognized object. The method further includes manipulating, using an object level augmentation model, a random property of the object by a random magnitude at an object level to determine a set of properties and a set of magnitudes of an object manipulation that maximizes a loss function of the image recognition model. The method also includes training a downstream task network based on a set of training data generated based on the set of properties and the set of magnitudes of the object manipulation, such that the loss function is minimized.

Abstract: A method for keypoint matching performed by a semantically aware keypoint matching model includes generating a semanticly segmented image from an image captured by a sensor of an agent, the semanticly segmented image associating a respective semantic label with each pixel of a group of pixels associated with the image. The method also includes generating a set of augmented keypoint descriptors by augmenting, for each keypoint of the set of keypoints associated with the image, a keypoint descriptor with semantic information associated with one or more pixels, of the semantically segmented image, corresponding to the keypoint. The method further includes controlling an action of the agent in accordance with identifying a target image having one or more first augmented keypoint descriptors that match one or more second augmented keypoint descriptors of the set of augmented keypoint descriptors.

Abstract: A method of representation learning for object detection from unlabeled point cloud sequences is described. The method includes detecting moving object traces from temporally-ordered, unlabeled point cloud sequences. The method also includes extracting a set of moving objects based on the moving object traces detected from the sequence of temporally-ordered, unlabeled point cloud sequences. The method further includes classifying the set of moving objects extracted from on the moving object traces detected from the sequence of temporally-ordered, unlabeled point cloud sequences. The method also includes estimating 3D bounding boxes for the set of moving objects based on the classifying of the set of moving objects.

Abstract: A method of conditional neural ground planes for static-dynamic disentanglement is described. The method includes extracting, using a convolutional neural network (CNN), CNN image features from an image to form a feature tensor. The method also includes resampling unprojected 2D features of the feature tensor to form feature pillars. The method further includes aggregating the feature pillars to form an entangled neural ground plane. The method also includes decomposing the entangled neural ground plane into a static neural ground plane and a dynamic neural ground plane.

Abstract: A method for multi-camera monocular depth estimation using pose averaging is described. The method includes determining a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes determining a multi-camera pose consistency constraint (PCC) loss associated with the multi-camera rig of the ego vehicle. The method further includes adjusting the multi-camera photometric loss according to the multi-camera PCC loss to form a multi-camera PCC photometric loss. The method also includes training a multi-camera depth estimation model and an ego-motion estimation model according to the multi-camera PCC photometric loss. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the trained multi-camera depth estimation model and the ego-motion estimation model.

Abstract: A method for depth estimation performed by a depth estimation system associated with an agent includes determining a first depth of a first image and a second depth of a second image, the first image and the second image being captured by a sensor associated with the agent. The method also includes generating a first 3D image of the first image based on the first depth, a first pose associated with the sensor, and the second image. The method further includes generating a warped depth image based on transforming the first depth in accordance with the first pose. The method also includes updating the first pose based on a second pose associated with the warped depth image and the second depth, and updating the first 3D image based on the updated first pose. The method further includes controlling an action of the agent based on the updated first 3D image.

Abstract: A method for 3D object detection is described. The method includes predicting, using a trained monocular depth network, an estimated monocular input depth map of a monocular image of a video stream and an estimated depth uncertainty map associated with the estimated monocular input depth map. The method also includes feeding back a depth uncertainty regression loss associated with the estimated monocular input depth map during training of the trained monocular depth network to update the estimated monocular input depth map. The method further includes detecting 3D objects from a 3D point cloud computed from the estimated monocular input depth map based on seed positions selected from the 3D point cloud and the estimated depth uncertainty map. The method also includes selecting 3D bounding boxes of the 3D objects detected from the 3D point cloud based on the seed positions and an aggregated depth uncertainty.

Abstract: A method for 3D object detection is described. The method includes concurrently training a monocular depth network and a 3D object detection network. The method also includes predicting, using a trained monocular depth network, a monocular depth map of a monocular image of a video stream. The method further includes inferring a 3D point cloud of a 3D object within the monocular image according to the predicted monocular depth map. The method also includes predicting 3D bounding boxes from a selection of 3D points from the 3D point cloud of the 3D object based on a selection regression loss.

Abstract: A method for 2D semantic keypoint detection and tracking is described. The method includes learning embedded descriptors of salient object keypoints detected in previous images according to a descriptor embedding space model. The method also includes predicting, using a shared image encoder backbone, salient object keypoints within a current image of a video stream. The method further includes inferring an object represented by the predicted, salient object keypoints within the current image of the video stream. The method also includes tracking the inferred object by matching embedded descriptors of the predicted, salient object keypoints representing the inferred object within the previous images of the video stream based on the descriptor embedding space model.

Abstract: A method for multi-camera monocular depth estimation using pose averaging is described. The method includes determining a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes determining a multi-camera pose consistency constraint (PCC) loss associated with the multi-camera rig of the ego vehicle. The method further includes adjusting the multi-camera photometric loss according to the multi-camera PCC loss to form a multi-camera PCC photometric loss. The method also includes training a multi-camera depth estimation model and an ego-motion estimation model according to the multi-camera PCC photometric loss. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the trained multi-camera depth estimation model and the ego-motion estimation model.

Abstract: A method for multi-camera self-supervised depth evaluation is described. The method includes training a self-supervised depth estimation model and an ego-motion estimation model according to a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes generating a single-scale correction factor according to a depth map of each camera of the multi-camera rig during a time-step. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the self-supervised depth estimation model and the ego-motion estimation model. The method also includes scaling the 360° point cloud according to the single-scale correction factor to form an aligned 360° point cloud.

Abstract: A method for scale-aware depth estimation using multi-camera projection loss is described. The method includes determining a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes training a scale-aware depth estimation model and an ego-motion estimation model according to the multi-camera photometric loss. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the scale-aware depth estimation model and the ego-motion estimation model. The method also includes planning a vehicle control action of the ego vehicle according to the 360° point cloud of the scene surrounding the ego vehicle.