Lowis3D: Language-Driven Open-World Instance-Level 3D Scene Understanding.

Open-world instance-level scene understanding aims to locate and recognize unseen object categories that are not present in the annotated dataset. This task is challenging because the model needs to both localize novel 3D objects and infer their semantic categories. A key factor for the recent progress in 2D open-world perception is the availability of large-scale image-text pairs from the Internet, which cover a wide range of vocabulary concepts. However, this success is hard to replicate in 3D scenarios due to the scarcity of 3D-text pairs. To address this challenge, we propose to harness pre-trained vision-language (VL) foundation models that encode extensive knowledge from image-text pairs to generate captions for multi-view images of 3D scenes. This allows us to establish explicit associations between 3D shapes and semantic-rich captions. Moreover, to enhance the fine-grained visual-semantic representation learning from captions for object-level categorization, we design hierarchical point-caption association methods to learn semantic-aware embeddings that exploit the 3D geometry between 3D points and multi-view images. In addition, to tackle the localization challenge for novel classes in the open-world setting, we develop debiased instance localization, which involves training object grouping modules on unlabeled data using instance-level pseudo supervision. This significantly improves the generalization capabilities of instance grouping and, thus, the ability to accurately locate novel objects. We conduct extensive experiments on 3D semantic, instance, and panoptic segmentation tasks, covering indoor and outdoor scenes across three datasets. Our method outperforms baseline methods by a significant margin in semantic segmentation (e.g., 34.5%∼65.3%), instance segmentation (e.g., 21.8%∼54.0%), and panoptic segmentation (e.g., 14.7%∼43.3%). Code will be available.

ST3D++: Denoised Self-Training for Unsupervised Domain Adaptation on 3D Object Detection.

In this paper, we present a self-training method, named ST3D++, with a holistic pseudo label denoising pipeline for unsupervised domain adaptation on 3D object detection. ST3D++ aims at reducing noise in pseudo label generation as well as alleviating the negative impacts of noisy pseudo labels on model training. First, ST3D++ pre-trains the 3D object detector on the labeled source domain with random object scaling (ROS) which is designed to reduce target domain pseudo label noise arising from object scale bias of the source domain. Then, the detector is progressively improved through alternating between generating pseudo labels and training the object detector with pseudo-labeled target domain data. Here, we equip the pseudo label generation process with a hybrid quality-aware triplet memory to improve the quality and stability of generated pseudo labels. Meanwhile, in the model training stage, we propose a source data assisted training strategy and a curriculum data augmentation policy to effectively rectify noisy gradient directions and avoid model over-fitting to noisy pseudo labeled data. These specific designs enable the detector to be trained on meticulously refined pseudo labeled target data with denoised training signals, and thus effectively facilitate adapting an object detector to a target domain without requiring annotations. Finally, our method is assessed on four 3D benchmark datasets (i.e., Waymo, KITTI, Lyft, and nuScenes) for three common categories (i.e., car, pedestrian and bicycle). ST3D++ achieves state-of-the-art performance on all evaluated settings, outperforming the corresponding baseline by a large margin (e.g., 9.6%  âˆ¼  38.16% on Waymo → KITTI in terms of AP[Formula: see text]), and even surpasses the fully supervised oracle results on the KITTI 3D object detection benchmark with target prior. Code is available at https://github.com/CVMI-Lab/ST3D.

Tuning C-C sp2/sp3 ratio of DLC films in FCVA system for biomedical application.

Diamond like carbon (DLC) films with different C-C sp2/sp3 ratios were prepared by tuning the N2 flow rate in a filtered cathodic vacuum arc (FCVA) system. The increase of N2 flow rate facilitated the increase of C-C sp2/sp3 ratio (1.09-2.66), the growth of particle size (0.78-1.58 nm) and the improvement of surface roughness. The SBF immersion results, as well as WCAs (77.57°~71.71°), hemolysis rate (0.14-1.00%) and cytotoxicity level (0) demonstrated that the as-fabricated DLC film was promising for biomedical application. As a result of surface charge effect, the apatite layers formed in the SBF increased with the increase of C-C sp2/sp3 ratio until 1.74 and then showed a tiny decrease during 1.74-2.66. A raise of hemolysis and cytotoxicity was observed when sp2/sp3 ratio was increased. Moreover, a decrease of friction coefficient of Si surface induced by increasing sp2/sp3 ratio was respectively evidenced in ambient air and SBF lubrication environments.

Replication and bioactivation of Ti-based alloy scaffold macroscopically identical to cancellous bone from polymeric template with TiNbZr powders.

In the present work, a new type of porous Ti-based alloy scaffold with high porosity (about 75%) and interconnected pores in the range of 300-1000 µm was fabricated by polymeric foam replication method with TiNbZr powders. This porous scaffold, which is consisted with major ß phase Ti and minor α Ti phase, exhibits a compressive strength of 14.9 MPa and an elastic modulus of 0.21 GPa, resembling the mechanical properties of nature human cancellous bone (σ = 10-50 MPa, E = 0.01-3.0 GPa). To improve its osteogenic potential, a bioactive nanostructural titanate network coating was applied to the scaffold surface using hydrothermal treatment. The bone-like apatite inducing ability of the treated scaffold was systemically assessed using SBF immersion during 3-28 days. The nanostructural titanate network coated on porous TiNbZr scaffold is favorable for apatite nucleation and subsequent growth due to the hydrolysis of titanate. The results suggest that highly porous TiNbZr scaffolds with an appropriate bioactive coating, which was fabricated in this study, could be potentially used for bone tissue engineering application.

Cloning and characterization of an antifungal class III chitinase from suspension-cultured bamboo ( Bambusa oldhamii ) cells.

A class III chitinase cDNA (BoChi3-1) was cloned using a cDNA library from suspension-cultured bamboo ( Bambusa oldhamii ) cells and then transformed into yeast ( Pichia pastoris X-33) for expression. Two recombinant chitinases with molecular masses of 28.3 and 35.7 kDa, respectively, were purified from the yeast's culture broth to electrophoretic homogeneity using sequential ammonium sulfate fractionation, Phenyl-Sepharose hydrophobic interaction chromatography, and Con A-Sepharose chromatography steps. N-Terminal sequencing and immunoblotting revealed that both recombinant chitinases were encoded by BoChi3-1, whereas SDS-PAGE and glycoprotein staining showed that the 35.7 kDa isoform (35.7 kDa BoCHI3-1) was glycosylated and the 28.3 kDa isoform (28.3 kDa BoCHI3-1) was not. For hydrolysis of ethylene glycol chitin (EGC), the optimal pH values were 3 and 4 for 35.7 and 28.3 kDa BoCHI3-1, respectively; the optimal temperatures were 80 and 70 degrees C, and the K(m) values were 1.35 and 0.65 mg/mL. The purified 35.7 kDa BoCHI3-1 hydrolyzed EGC more efficiently than the 28.3 kDa isoform, as compared with their specific activity and activation energy. Both recombinant BoCHI3-1 isoforms showed antifungal activity against Scolecobasidium longiphorum and displayed remarkable thermal (up to 70 degrees C) and storage (up to a year at 4 degrees C) stabilities.