Adaptive Masked Weight Imprinting for Few-Shot Segmentation
By Mennatullah Siam, Boris N. Oreshkin. 2019
35 citations recorded by the source corpus.
Deep learning has mainly thrived by training on large-scale datasets. However, for a continual learning agent it is critical to incrementally update its model in a sample efficient manner. Learning semantic segmentation from few labelled samples can be a significant step toward such goal. We propose a novel method that constructs the new class weights from few labelled samples in the support set without back-propagation, while updating the previously learned classes. Inspiring from the work on adaptive correlation filters, an adaptive masked imprinted weights method is designed. It utilizes a masked average pooling layer on the output embeddings and acts as a positive proxy for that class. Our proposed method is evaluated on PASCAL-5i dataset and outperforms the state of the art in the 5-shot semantic segmentation. Unlike previous methods, our proposed approach does not require a second branch to estimate parameters or prototypes, and enables the adaptation of previously learned weights. Our adaptation scheme is evaluated on DAVIS video segmentation benchmark and our proposed incremental version of PASCAL and has shown to outperform the baseline model.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Guosheng Lin
- Chi Zhang
- Stefano Soatto
- Alessandro Achille
- Avinash Ravichandran
- Han-Jia Ye
- Rahul Bhotika
- Qing Liu
In Computer Vision and Pattern Recognition
- Guosheng Lin
- Chi Zhang
- Stefano Soatto
- Chunhua Shen
- Luc Van Gool
- Bernt Schiele
- Dacheng Tao
- Trevor Darrell
In Radiology, Nuclear Medicine and Imaging
- Chi Zhang
- Guosheng Lin
- Rui Yao
- Alessandro Achille
- Stefano Soatto
- Songyang Zhang
- Yongfei Liu
- Xiangyi Zhang
Citations in the corpus
- Incremental Meta-Learning via Indirect Discriminant Alignment (2020, background): “[40, 34] extended this class-incremental framework to other visual recognition tasks like semantic segmentation”
- CRNet: Cross-Reference Networks for Few-Shot Segmentation (2020, unknown): “1 R-DRCN [18] VGG16 40.”
- Pyramid Graph Networks With Connection Attentions for Region-Based One-Shot Semantic Segmentation (2019, background): “Model (5-shot) fold-0 fold-1 fold-2 fold-3 mean co-FCN [15] 37.5 50.0 44.1 33.9 41.4 OSLSM [18] 35.9 58.1 42.7 39.1 43.9 Reduced-DFCN8s [19] 45.3 51.4 44.9 39.5 45.3 SG-One [27] 41.9 (a) Evaluation metric of mean class-wise IoU adopted in [18] Model (5-shot) IoU co-FCN [15] 60.2 OSLSM [18] 61.5…”
- Prior Guided Feature Enrichment Network for Few-Shot Segmentation (2020, unknown): “Existing few-shot segmentation frameworks [4], [13], [24], [26], [29], [31], [40], [46] use masked global average pooling for extracting class vectors from support images before further processing.”
- Part-aware Prototype Network for Few-shot Semantic Segmentation (2020, methodology): “Second, other prototype-based methods [7,27,33] typically use a holistic representation for each semantic class, which is difficult to cope with diverse appearance in objects with different parts, poses or subcategories.”