Unlike closed-vocabulary 3D instance segmentation that is often trained end-to-end, open-vocabulary 3D instance segmentation (OV-3DIS) often leverages vision-language models (VLMs) to generate 3D instance proposals and classify them. While various concepts have been proposed from existing research, we observe that these individual concepts are not mutually exclusive but complementary. In this paper, we propose a new state-of-the-art solution for OV-3DIS by carefully designing a recipe to combine the concepts together and refining them to address key challenges. Our solution follows the two-stage scheme: 3D proposal generation and instance classification. We employ robust 3D tracking-based proposal aggregation to generate 3D proposals and remove overlapped or partial proposals by iterative merging/removal. For the classification stage, we replace the standard CLIP model with Alpha-CLIP, which incorporates object masks as an alpha channel to reduce background noise and obtain object-centric representation. Additionally, we introduce the standardized maximum similarity (SMS) score to normalize text-to-proposal similarity, effectively filtering out false positives and boosting precision. Our framework achieves state-of-the-art performance on ScanNet200 and S3DIS across all AP and AR metrics, even surpassing an end-to-end closed-vocabulary method.

The field of indoor monocular 3D object detection is gaining significant attention, fueled by the increasing demand in VR/AR and robotic applications. However, its advancement is impeded by the limited availability and diversity of 3D training data, owing to the labor-intensive nature of 3D data collection and annotation processes. In this paper, we present V-MIND (Versatile Monocular INdoor Detector), which enhances the performance of in-door 3D detectors across a diverse set of object classes by harnessing publicly available large-scale 2D datasets. By leveraging well-established monocular depth estimation techniques and camera intrinsic predictors, we can generate 3D training data by converting large-scale 2D images into 3D point clouds and subsequently deriving pseudo 3D bounding boxes. To mitigate distance errors inherent in the converted point clouds, we introduce a novel 3D self-calibration loss for refining the pseudo 3D bounding boxes during training. Additionally, we propose a novel ambiguity loss to address the ambiguity that arises when introducing new classes from 2D datasets. Finally, through joint training with existing 3D datasets and pseudo 3D bounding boxes derived from 2D datasets, V-MIND achieves state-of-the-art object detection performance across a wide range of classes on the Omni3D indoor dataset.

Large-scale pre-trained vision-language models (VLM) such as CLIP [32] have demonstrated noteworthy zero-shot classification capability, achieving 76.3% top-1 accuracy on ImageNet without seeing any examples. However, while applying CLIP to a downstream target domain, the presence of visual and text domain gaps and cross-modality misalignment can greatly impact the model performance. To address such challenges, we propose ReCLIP, a novel source-free domain adaptation method for VLMs, which does not require any source data or target labeled data. ReCLIP first learns a projection space to mitigate the misaligned visual-text embeddings and learns pseudo labels. Then, it deploys cross-modality self-training with the pseudo labels to update visual and text encoders, refine labels and reduce domain gaps and misalignment iteratively. With extensive experiments, we show that ReCLIP outperforms all the baselines significantly and improves the average accuracy of CLIP from 69.83% to 74.94% on 22 image classification benchmarks.

We propose ImGeoNet, a multi-view image-based 3D object detection framework that models a 3D space by an image-induced geometry-aware voxel representation. Unlike previous methods which aggregate 2D features into 3D voxels without considering geometry, ImGeoNet learns to induce geometry from multi-view images to alleviate the confusion arising from voxels of free space, and during the inference phase, only images from multiple views are required. Besides, a powerful pre-trained 2D feature extractor can be leveraged by our representation, leading to a more robust performance. To evaluate the effectiveness of ImGeoNet, we conduct quantitative and qualitative experiments on three indoor datasets, namely ARKitScenes, ScanNetV2, and ScanNet200. The results demonstrate that ImGeoNet outperforms the current state-of-the-art multiview image-based method, ImVoxelNet, on all three datasets in terms of detection accuracy. In addition, ImGeoNet shows great data efficiency by achieving results comparable to ImVoxelNet with 100 views while utilizing only 40 views. Furthermore, our studies indicate that our proposed image-induced geometry-aware representation can enable image-based methods to attain superior detection accuracy than the seminal point cloud-based method, VoteNet, in two practical scenarios: (1) scenarios where point clouds are sparse and noisy, such as in ARKitScenes, and (2) scenarios involve diverse object classes, particularly classes of small objects, as in the case in ScanNet200. Project page: https://ttaoretw.github.io/imgeonet.

Accurate per-pixel semantic class annotations of the entire video are crucial for designing and evaluating video semantic segmentation algorithms. However, the annotations are usually limited to a small subset of the video frames due to the high annotation cost and limited budget in practice. In this paper, we propose a novel human-in-the-loop framework called HVSA to generate semantic segmentation annotations for the entire video using only a small annotation budget. Our method alternates between active sample selection and test-time fine-tuning algorithms until annotation quality is satisfied. In particular, the active sample selection algorithm picks the most important samples to get manual annotations, where the sample can be a video frame, a rectangle, or even a super-pixel. Further, the test-time fine-tuning algorithm propagates the manual annotations of selected samples to the entire video. Real-world experiments show that our method generates highly accurate and consistent semantic segmentation annotations while simultaneously enjoys significantly small annotation cost.

To improve the generalization of 3D human pose estimators, many existing deep learning based models focus on adding different augmentations to training poses. However, data augmentation techniques are limited to the "seen" pose combinations and hard to infer poses with rare "unseen" joint positions. To address this problem, we present CameraPose, a weakly-supervised framework for 3D human pose estimation from a single image, which can not only be applied on 2D-3D pose pairs but also on 2D alone annotations. By adding a camera parameter branch, any in-the-wild 2D annotations can be fed into our pipeline to boost the training diversity and the 3D poses can be implicitly learned by reprojecting back to 2D. Moreover, CameraPose introduces a refinement network module with confidence-guided loss to further improve the quality of noisy 2D keypoints extracted by 2D pose estimators. Experimental results demonstrate that the CameraPose brings in clear improvements on cross-scenario datasets. Notably, it outperforms the baseline method by 3mm on the most challenging dataset 3DPW. In addition, by combining our proposed refinement network module with existing 3D pose estimators, their performance can be improved in cross-scenario evaluation.

Video summarization has unprecedented importance to help us digest, browse, and search today's ever-growing video collections. We propose a novel subset selection technique that leverages supervision in the form of humancreated summaries to perform automatic keyframe-based video summarization. The main idea is to nonparametrically transfer summary structures from annotated videos to unseen test videos. We show how to extend our method to exploit semantic side information about the video's category/ genre to guide the transfer process by those training videos semantically consistent with the test input. We also show how to generalize our method to subshot-based summarization, which not only reduces computational costs but also provides more flexible ways of defining visual similarity across subshots spanning several frames. We conduct extensive evaluation on several benchmarks and demonstrate promising results, outperforming existing methods in several settings.

In this paper we propose a novel method for image semantic segmentation using multiple graphs. The multiview affinity graph is constructed by leveraging the consistency between semantic space and multiple visualspaces. With block-diagonal constraints, we enforce the affinity matrix to be sparse such that the pairwise potential for dissimilar superpixels is close to zero. By a divide-and-conquer strategy, the optimizationfor learning affinity matrix is decomposed into several subproblems that can be solved in parallel. Using the neighborhood relationship between superpixels and the consistency between affinity matrix and labelconfidencematrix, we infer the semantic label for each superpixel of unlabeled images by minimizing an objective whose closed form solution can be easily obtained. Experimental results on two real-world image datasetsdemonstrate the effectiveness of our method.

There are a large number of images available on the web; meanwhile, only a subset of web images can be labeled by professionals because manual annotation is time-consuming and labor-intensive. Although we can now use the collaborative image tagging system, e.g., Flickr, to get a lot of tagged images provided by Internet users, these labels may be incorrect or incomplete. Furthermore, semantics richness requires more than one label to describe one image in real applications, and multiple labels usually interact with each other in semantic space. It is of significance to learn semantic context with large-scale weakly-labeled image set in the task of multi-label annotation. In this paper, we develop a novel method to learn semantic context and predict the labels of web images in a semi-supervised framework. To address the scalability issue, a small number of exemplar images are first obtained to cover the whole data cloud; then the label vector of each image is estimated as a local combination of the exemplar label vectors. Visual context, semantic context, and neighborhood consistency in both visual and semantic spaces are sufficiently leveraged in the proposed framework. Finally, the semantic context and the label confidence vectors for exemplar images are both learned in an iterative way. Experimental results on the real-world image dataset demonstrate the effectiveness of our method.