Data exploitation: multi-task learning of object detection and semantic segmentation on partially annotated data

Hoàng-Ân Lê, Minh-Tan Pham
11/7/2023
cs.CV

Official code: lhoangan/multas (8 stars, last commit 2025-08-06).

Abstract

Multi-task partially annotated data where each data point is annotated for only a single task are potentially helpful for data scarcity if a network can leverage the inter-task relationship. In this paper, we study the joint learning of object detection and semantic segmentation, the two most popular vision problems, from multi-task data with partial annotations. Extensive experiments are performed to evaluate each task performance and explore their complementarity when a multi-task network cannot optimize both tasks simultaneously. We propose employing knowledge distillation to leverage joint-task optimization. The experimental results show favorable results for multi-task learning and knowledge distillation over single-task learning and even full supervision scenario. All code and data splits are available at https://github.com/lhoangan/multas

View on arXivView PDF

Code Implementations(1)

lhoangan/multasOfficial100%

Multi-Task Self-training for Object detection and Semantic segmentation

81Shell, PythonAug 9, 20231 years ago

Similar Papers

Multi-Task Consistency for Active Learning

88% match2023

Box for Mask and Mask for Box: weak losses for multi-task partially supervised learning

88% match1 repo82024

Multi-Task Partially Supervised Learning for Super-Resolution and Semantic Segmentation on Earth Observation data

87% match2026

Leveraging knowledge distillation for partial multi-task learning from multiple remote sensing datasets

86% match1 repo82024

YOLOR-Based Multi-Task Learning

86% match1 repo2.0k2023

Multi-Task Self-Supervised Learning for Image Segmentation Task

85% match2023

Cite this paper

@article{l2023data,
  title  = {Data exploitation: multi-task learning of object detection and semantic segmentation on partially annotated data},
  author = {Hoàng-Ân Lê and Minh-Tan Pham},
  year   = {2023},
  eprint = {2311.04040},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2311.04040v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2311.04040.svg)](https://codeofpaper.com/paper/2311.04040)

Discussion