SD-Net: Symmetric-Aware Keypoint Prediction and Domain Adaptation for 6D Pose Estimation In Bin-picking Scenarios

Despite the success in 6D pose estimation in bin-picking scenarios, existing methods still struggle to produce accurate prediction results for symmetry objects and real world scenarios. The primary bottlenecks include 1) the ambiguity keypoints caused by object symmetries; 2) the domain gap between real and synthetic data. To circumvent these problem, we propose a new 6D pose estimation network with symmetric-aware keypoint prediction and self-training domain adaptation (SD-Net). SD-Net builds on pointwise keypoint regression and deep hough voting to perform reliable detection keypoint under clutter and occlusion. Specifically, at the keypoint prediction stage, we designe a robust 3D keypoints selection strategy considering the symmetry class of objects and equivalent keypoints, which facilitate locating 3D keypoints even in highly occluded scenes. Additionally, we build an effective filtering algorithm on predicted keypoint to dynamically eliminate multiple ambiguity and outlier keypoint candidates. At the domain adaptation stage, we propose the self-training framework using a student-teacher training scheme. To carefully distinguish reliable predictions, we harnesses a tailored heuristics for 3D geometry pseudo labelling based on semi-chamfer distance. On public Sil'eane dataset, SD-Net achieves state-of-the-art results, obtaining an average precision of 96%. Testing learning and generalization abilities on public Parametric datasets, SD-Net is 8% higher than the state-of-the-art method. The code is available at https://github.com/dingthuang/SD-Net.

NormNet: Scale Normalization for 6D Pose Estimation in Stacked Scenarios

Existing Object Pose Estimation (OPE) methods for stacked scenarios are not robust to changes in object scale. This paper proposes a new 6DoF OPE network (NormNet) for different scale objects in stacked scenarios. Specifically, each object's scale is first learned with point-wise regression. Then, all objects in the stacked scenario are normalized into the same scale through semantic segmentation and affine transformation. Finally, they are fed into a shared pose estimator to recover their 6D poses. In addition, we introduce a new Sim-to-Real transfer pipeline, combining style transfer and domain randomization. This improves the NormNet's performance on real data even if we only train it on synthetic data. Extensive experiments demonstrate that the proposed method achieves state-of-the-art performance on public benchmarks and the MultiScale dataset we constructed. The real-world experiments show that our method can robustly estimate the 6D pose of objects at different scales.

Medical Image Segmentation via Sparse Coding Decoder

Transformers have achieved significant success in medical image segmentation, owing to its capability to capture long-range dependencies. Previous works incorporate convolutional layers into the encoder module of transformers, thereby enhancing their ability to learn local relationships among pixels. However, transformers may suffer from limited generalization capabilities and reduced robustness, attributed to the insufficient spatial recovery ability of their decoders. To address this issue, A convolution sparse vector coding based decoder is proposed , namely CAScaded multi-layer Convolutional Sparse vector Coding DEcoder (CASCSCDE), which represents features extracted by the encoder using sparse vectors. To prove the effectiveness of our CASCSCDE, The widely-used TransUNet model is chosen for the demonstration purpose, and the CASCSCDE is incorporated with TransUNet to establish the TransCASCSCDE architecture. Our experiments demonstrate that TransUNet with CASCSCDE significantly enhances performance on the Synapse benchmark, obtaining up to 3.15\% and 1.16\% improvements in DICE and mIoU scores, respectively. CASCSCDE opens new ways for constructing decoders based on convolutional sparse vector coding.

GRID: Scene-Graph-based Instruction-driven Robotic Task Planning

Recent works have shown that Large Language Models (LLMs) can promote grounding instructions to robotic task planning. Despite the progress, most existing works focused on utilizing raw images to help LLMs understand environmental information, which not only limits the observation scope but also typically requires massive multimodal data collection and large-scale models. In this paper, we propose a novel approach called Graph-based Robotic Instruction Decomposer (GRID), leverages scene graph instead of image to perceive global scene information and continuously plans subtask in each stage for a given instruction. Our method encodes object attributes and relationships in graphs through an LLM and Graph Attention Networks, integrating instruction features to predict subtasks consisting of pre-defined robot actions and target objects in the scene graph. This strategy enables robots to acquire semantic knowledge widely observed in the environment from the scene graph. To train and evaluate GRID, we build a dataset construction pipeline to generate synthetic datasets in graph-based robotic task planning. Experiments have shown that our method outperforms GPT-4 by over 25.4% in subtask accuracy and 43.6% in task accuracy. Experiments conducted on datasets of unseen scenes and scenes with different numbers of objects showed that the task accuracy of GRID declined by at most 3.8%, which demonstrates its good cross-scene generalization ability. We validate our method in both physical simulation and the real world.

Reinforcement Learning Based Pushing and Grasping Objects from Ungraspable Poses

Grasping an object when it is in an ungraspable pose is a challenging task, such as books or other large flat objects placed horizontally on a table. Inspired by human manipulation, we address this problem by pushing the object to the edge of the table and then grasping it from the hanging part. In this paper, we develop a model-free Deep Reinforcement Learning framework to synergize pushing and grasping actions. We first pre-train a Variational Autoencoder to extract high-dimensional features of input scenario images. One Proximal Policy Optimization algorithm with the common reward and sharing layers of Actor-Critic is employed to learn both pushing and grasping actions with high data efficiency. Experiments show that our one network policy can converge 2.5 times faster than the policy using two parallel networks. Moreover, the experiments on unseen objects show that our policy can generalize to the challenging case of objects with curved surfaces and off-center irregularly shaped objects. Lastly, our policy can be transferred to a real robot without fine-tuning by using CycleGAN for domain adaption and outperforms the push-to-wall baseline.

3DFill:Reference-guided Image Inpainting by Self-supervised 3D Image Alignment

Most existing image inpainting algorithms are based on a single view, struggling with large holes or the holes containing complicated scenes. Some reference-guided algorithms fill the hole by referring to another viewpoint image and use 2D image alignment. Due to the camera imaging process, simple 2D transformation is difficult to achieve a satisfactory result. In this paper, we propose 3DFill, a simple and efficient method for reference-guided image inpainting. Given a target image with arbitrary hole regions and a reference image from another viewpoint, the 3DFill first aligns the two images by a two-stage method: 3D projection + 2D transformation, which has better results than 2D image alignment. The 3D projection is an overall alignment between images and the 2D transformation is a local alignment focused on the hole region. The entire process of image alignment is self-supervised. We then fill the hole in the target image with the contents of the aligned image. Finally, we use a conditional generation network to refine the filled image to obtain the inpainting result. 3DFill achieves state-of-the-art performance on image inpainting across a variety of wide view shifts and has a faster inference speed than other inpainting models.

KRF: Keypoint Refinement with Fusion Network for 6D Pose Estimation

Existing refinement methods gradually lose their ability to further improve pose estimation methods' accuracy. In this paper, we propose a new refinement pipeline, Keypoint Refinement with Fusion Network (KRF), for 6D pose estimation, especially for objects with serious occlusion. The pipeline consists of two steps. It first completes the input point clouds via a novel point completion network. The network uses both local and global features, considering the pose information during point completion. Then, it registers the completed object point cloud with corresponding target point cloud by Color supported Iterative KeyPoint (CIKP). The CIKP method introduces color information into registration and registers point cloud around each keypoint to increase stability. The KRF pipeline can be integrated with existing popular 6D pose estimation methods, e.g. the full flow bidirectional fusion network, to further improved their pose estimation accuracy. Experiments show that our method outperforms the state-of-the-art method from 93.9\% to 94.4\% on YCB-Video dataset and from 64.4\% to 66.8\% on Occlusion LineMOD dataset. Our source code is available at https://github.com/zhanhz/KRF.

Class-Aware Contrastive Semi-Supervised Learning

Pseudo-label-based semi-supervised learning (SSL) has achieved great success on raw data utilization. However, its training procedure suffers from confirmation bias due to the noise contained in self-generated artificial labels. Moreover, the model's judgment becomes noisier in real-world applications with extensive out-of-distribution data. To address this issue, we propose a general method named Class-aware Contrastive Semi-Supervised Learning (CCSSL), which is a drop-in helper to improve the pseudo-label quality and enhance the model's robustness in the real-world setting. Rather than treating real-world data as a union set, our method separately handles reliable in-distribution data with class-wise clustering for blending into downstream tasks and noisy out-of-distribution data with image-wise contrastive for better generalization. Furthermore, by applying target re-weighting, we successfully emphasize clean label learning and simultaneously reduce noisy label learning. Despite its simplicity, our proposed CCSSL has significant performance improvements over the state-of-the-art SSL methods on the standard datasets CIFAR100 and STL10. On the real-world dataset Semi-iNat 2021, we improve FixMatch by 9.80% and CoMatch by 3.18%. Code is available https://github.com/TencentYoutuResearch/Classification-SemiCLS.

SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filters

Deep neural networks have enjoyed remarkable success for various vision tasks, however it remains challenging to apply CNNs to domains lacking a regular underlying structures such as 3D point clouds. Towards this we propose a novel convolutional architecture, termed SpiderCNN, to efficiently extract geometric features from point clouds. SpiderCNN is comprised of units called SpiderConv, which extend convolutional operations from regular grids to irregular point sets that can be embedded in R^n, by parametrizing a family of convolutional filters. We design the filter as a product of a simple step function that captures local geodesic information and a Taylor polynomial that ensures the expressiveness. SpiderCNN inherits the multi-scale hierarchical architecture from classical CNNs, which allows it to extract semantic deep features. Experiments on ModelNet40 demonstrate that SpiderCNN achieves state-of-the-art accuracy 92.4% on standard benchmarks, and shows competitive performance on segmentation task.