Joint Task-Recursive Learning for Semantic Segmentation and Depth Estimation
By Zhenyu Zhang, Zhen Cui, Chunyan Xu, Zequn Jie, Xiang-Yang Li, Jian Yang. 2018
234 citations recorded by the source corpus.
Subfields: Computer Vision and Pattern Recognition, Media Technology
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Xiaogang Wang
- Yi Yang
- Wanli Ouyang
- Ming-Hsuan Yang
- Hongsheng Li
- Dacheng Tao
- Chunhua Shen
In Media Technology
- Luc Van Gool
- Stamatios Georgoulis
- Jian Yang
- Simon Vandenhende
- Wei Wu
- Bert De Brabandere
- Nicu Sebe
- Wanli Ouyang
Citations in the corpus
- Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference (2020, methodology): “mphasize learning feature representations from multi-task supervisory signals by employing architectures that encode shared and task-specic information. On the other hand, decoder-focused approaches [63,65,66,62] utilize the multi-task feature representations learned at the encoding stage to distill”
- Pattern-Affinitive Propagation Across Depth, Surface Normal and Semantic Segmentation (2019, methodology): “We randomly select more images (12k, same as [29, 62] ) from the raw data of official training scenes.”
- Backbone is All Your Need: A Simplified Architecture for Visual Object Tracking (2022, background): “, image classification [18, 13], object detection [33], semantic segmentation [17, 47], depth estimation [23, 37], etc.”
- Monocular depth estimation based on deep learning: An overview (2020, methodology): “[56] 2018 RGB + Depth √ Task-attentional, BerHu loss Xu et al.”
- Categorical Depth Distribution Network for Monocular 3D Object Detection (2021, methodology): “Further, many methods jointly perform depth estimation and segmentation [63, 66, 58, 14] in an end-to-end manner.”
- Branched Multi-Task Networks: Deciding what layers to share (2019, methodology): “nd using half versus the complete training set. The task clustering procedure resulted in the Method S (IoU) D (m) C (%) #P Single Task 41:01 0:344 63:31 125 RefineNet [30] 45:7 - - 118 Task-Recursive [59] 46:3 0:219 - - MTL (ours) 40:65 0:349 66:75 79 Table 4: Results on the SUN RGB-D test set. Our ”
- Learning Semantic Segmentation From Synthetic Data: A Geometrically Guided Input-Output Adaptation Approach (2018, methodology): “Besides, various techniques have been used for the task, including fine-tuning [40], cross-modality influence [23], task distillation module with intermediate auxiliary tasks [55], recursive estimation [58], task attention loss [24].”