skip to main content

Explore Research Products in the NSF-PAR

Title: UCTNet: Uncertainty-Aware Cross-Modal Transformer Network for Indoor RGB-D Semantic Segmentation
In this paper, we tackle the problem of RGB-D Semantic Segmentation. The key challenges in solving this problem lie in 1) how to extract features from depth sensor data and 2) how to effectively fuse the features extracted from the two modalities. For the first challenge, we found that the depth information obtained from the sensor is not always reliable (e.g. objects with reflective or dark surfaces typically have inaccurate or void sensor readings), and existing methods that extract depth features using ConvNets did not explicitly consider the reliability of depth value at different pixel locations. To tackle this challenge, we propose a novel mechanism, namely Uncertainty-Aware Self-Attention that explicitly controls the information flow from unreliable depth pixels to confident depth pixels during feature extraction. For the second challenge, we propose an effective and scalable fusion module based on Cross-Attention that can adaptively fuse and exchange information between the RGB encoder and depth encoder. Our proposed framework, namely UCTNet, is an encoder-decoder network that naturally incorporates these two key designs for robust and accurate RGB-D Segmentation. Experimental results show that UCTNet outperforms existing works and achieves state-of-the-art performances on two RGB-D Semantic Segmentation benchmarks.  more » « less
Award ID(s):
1931867
NSF-PAR ID:
10467720
Author(s) / Creator(s):
;
Editor(s):
Avidan, S.
Publisher / Repository:
Springer
Date Published:
Volume:
13690
Page Range / eLocation ID:
20-37
Subject(s) / Keyword(s):
["RGBD semantic segmentation","uncertainty aware","cross modal","transformer"]
Format(s):
Medium: X
Location:
Israel
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ( , IEEE Winter Conference on Applications of Computer Vision)
    null (Ed.)
    Training a semantic segmentation model requires large densely-annotated image datasets that are costly to obtain. Once the training is done, it is also difficult to add new object categories to such segmentation models. In this paper, we tackle the few-shot semantic segmentation problem, which aims to perform image segmentation task on unseen object categories merely based on one or a few support example(s). The key to solving this few-shot segmentation problem lies in effectively utilizing object information from support examples to separate target objects from the background in a query image. While existing methods typically generate object-level representations by averaging local features in support images, we demonstrate that such object representations are typically noisy and less distinguishing. To solve this problem, we design an object representation generator (ORG) module which can effectively aggregate local object features from support image( s) and produce better object-level representation. The ORG module can be embedded into the network and trained end-to-end in a weakly-supervised fashion without extra human annotation. We incorporate this design into a modified encoder-decoder network to present a powerful and efficient framework for few-shot semantic segmentation. Experimental results on the Pascal-VOC and MS-COCO datasets show that our approach achieves better performance compared to existing methods under both one-shot and five-shot settings. 
    more » « less
  2. ; ; ; ; ( , Diagnostics)
    Medical image segmentation is one of the most challenging tasks in medical image analysis and widely developed for many clinical applications. While deep learning-based approaches have achieved impressive performance in semantic segmentation, they are limited to pixel-wise settings with imbalanced-class data problems and weak boundary object segmentation in medical images. In this paper, we tackle those limitations by developing a new two-branch deep network architecture which takes both higher level features and lower level features into account. The first branch extracts higher level feature as region information by a common encoder-decoder network structure such as Unet and FCN, whereas the second branch focuses on lower level features as support information around the boundary and processes in parallel to the first branch. Our key contribution is the second branch named Narrow Band Active Contour (NB-AC) attention model which treats the object contour as a hyperplane and all data inside a narrow band as support information that influences the position and orientation of the hyperplane. Our proposed NB-AC attention model incorporates the contour length with the region energy involving a fixed-width band around the curve or surface. The proposed network loss contains two fitting terms: (i) a high level feature (i.e., region) fitting term from the first branch; (ii) a lower level feature (i.e., contour) fitting term from the second branch including the (ii1) length of the object contour and (ii2) regional energy functional formed by the homogeneity criterion of both the inner band and outer band neighboring the evolving curve or surface. The proposed NB-AC loss can be incorporated into both 2D and 3D deep network architectures. The proposed network has been evaluated on different challenging medical image datasets, including DRIVE, iSeg17, MRBrainS18 and Brats18. The experimental results have shown that the proposed NB-AC loss outperforms other mainstream loss functions: Cross Entropy, Dice, Focal on two common segmentation frameworks Unet and FCN. Our 3D network which is built upon the proposed NB-AC loss and 3DUnet framework achieved state-of-the-art results on multiple volumetric datasets. 
    more » « less
  3. ( , IEEE/CVF Winter Conference on Applications of Computer Vision)
    Transformers have shown great promise in medical image segmentation due to their ability to capture long-range dependencies through self-attention. However, they lack the ability to learn the local (contextual) relations among pixels. Previous works try to overcome this problem by embedding convolutional layers either in the encoder or decoder modules of transformers thus ending up sometimes with inconsistent features. To address this issue, we propose a novel attention-based decoder, namely CASCaded Attention DEcoder (CASCADE), which leverages the multiscale features of hierarchical vision transformers. CASCADE consists of i) an attention gate which fuses features with skip connections and ii) a convolutional attention module that enhances the long-range and local context by suppressing background information. We use a multi-stage feature and loss aggregation framework due to their faster convergence and better performance. Our experiments demonstrate that transformers with CASCADE significantly outperform state-of-the-art CNN- and transformer-based approaches, obtaining up to 5.07% and 6.16% improvements in DICE and mIoU scores, respectively. CASCADE opens new ways of designing better attention-based decoders. 
    more » « less
  4. ; ; ( , IEEE Winter Conference on Applications of Computer Vision)
    null (Ed.)
    Training a semantic segmentation model requires large densely-annotated image datasets that are costly to obtain. Once the training is done, it is also difficult to add new ob- ject categories to such segmentation models. In this pa- per, we tackle the few-shot semantic segmentation prob- lem, which aims to perform image segmentation task on un- seen object categories merely based on one or a few sup- port example(s). The key to solving this few-shot segmen- tation problem lies in effectively utilizing object informa- tion from support examples to separate target objects from the background in a query image. While existing meth- ods typically generate object-level representations by av- eraging local features in support images, we demonstrate that such object representations are typically noisy and less distinguishing. To solve this problem, we design an ob- ject representation generator (ORG) module which can ef- fectively aggregate local object features from support im- age(s) and produce better object-level representation. The ORG module can be embedded into the network and trained end-to-end in a weakly-supervised fashion without extra hu- man annotation. We incorporate this design into a modified encoder-decoder network to present a powerful and efficient framework for few-shot semantic segmentation. Experimen- tal results on the Pascal-VOC and MS-COCO datasets show that our approach achieves better performance compared to existing methods under both one-shot and five-shot settings. 
    more » « less
  5. ; ; ( , Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies)

    Learning the human--mobility interaction (HMI) on interactive scenes (e.g., how a vehicle turns at an intersection in response to traffic lights and other oncoming vehicles) can enhance the safety, efficiency, and resilience of smart mobility systems (e.g., autonomous vehicles) and many other ubiquitous computing applications. Towards the ubiquitous and understandable HMI learning, this paper considers both spoken language (e.g., human textual annotations) and unspoken language (e.g., visual and sensor-based behavioral mobility information related to the HMI scenes) in terms of information modalities from the real-world HMI scenarios. We aim to extract the important but possibly implicit HMI concepts (as the named entities) from the textual annotations (provided by human annotators) through a novel human language and sensor data co-learning design.

    To this end, we propose CG-HMI, a novel Cross-modality Graph fusion approach for extracting important Human-Mobility Interaction concepts from co-learning of textual annotations as well as the visual and behavioral sensor data. In order to fuse both unspoken and spoken languages, we have designed a unified representation called the human--mobility interaction graph (HMIG) for each modality related to the HMI scenes, i.e., textual annotations, visual video frames, and behavioral sensor time-series (e.g., from the on-board or smartphone inertial measurement units). The nodes of the HMIG in these modalities correspond to the textual words (tokenized for ease of processing) related to HMI concepts, the detected traffic participant/environment categories, and the vehicle maneuver behavior types determined from the behavioral sensor time-series. To extract the inter- and intra-modality semantic correspondences and interactions in the HMIG, we have designed a novel graph interaction fusion approach with differentiable pooling-based graph attention. The resulting graph embeddings are then processed to identify and retrieve the HMI concepts within the annotations, which can benefit the downstream human-computer interaction and ubiquitous computing applications. We have developed and implemented CG-HMI into a system prototype, and performed extensive studies upon three real-world HMI datasets (two on car driving and the third one on e-scooter riding). We have corroborated the excellent performance (on average 13.11% higher accuracy than the other baselines in terms of precision, recall, and F1 measure) and effectiveness of CG-HMI in recognizing and extracting the important HMI concepts through cross-modality learning. Our CG-HMI studies also provide real-world implications (e.g., road safety and driving behaviors) about the interactions between the drivers and other traffic participants.

     
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email