Towards Resolving the Challenge of Long-tail Distribution in UAV Images for Object Detection

  title={Towards Resolving the Challenge of Long-tail Distribution in UAV Images for Object Detection},
  author={Weiping Yu and Taojiannan Yang and Chen Chen},
  journal={2021 IEEE Winter Conference on Applications of Computer Vision (WACV)},
  • Weiping Yu, Taojiannan Yang, Chen Chen
  • Published 7 November 2020
  • Environmental Science, Computer Science
  • 2021 IEEE Winter Conference on Applications of Computer Vision (WACV)
Existing methods for object detection in UAV images ignored an important challenge – imbalanced class distribution in UAV images – which leads to poor performance on tail classes. We systematically investigate existing solutions to long-tail problems and unveil that re-balancing methods that are effective on natural image datasets cannot be trivially applied to UAV datasets. To this end, we rethink longtailed object detection in UAV images and propose the Dual Sampler and Head detection Network… 

Figures and Tables from this paper

VistrongerDet: Stronger Visual Information for Object Detection in VisDrone Images
A novel detector named VistrongerDet, which possesses Stronger Visual Information, is proposed, which integrates the novel components of FPN level, ROI level and head level enhancements and significantly improves the detection performance.
VisDrone-DET2021: The Vision Meets Drone Object detection Challenge Results
The VisDrone team gathered a massive data set and organized Vision Meets Drones: A Challenge (VisDrone2021) in conjunction with the IEEE International Conference on Computer Vision (ICCV 2021) to advance the field.
A Survey on Long-Tailed Visual Recognition
This survey focuses on the problems caused by long-tailed data distribution, sort out the representative long-tails visual recognition datasets and summarize some mainstream long-tail studies, and quantitatively study 20 widely-used and large-scale visual datasets proposed in the last decade.
Deep Long-Tailed Learning: A Survey
A comprehensive survey on recent advances in deep long-tailed learning is provided, highlighting important applications of deepLongtailed learning and identifying several promising directions for future research.
Towards Better Object Detection in Scale Variation with Adaptive Feature Selection
A novel adaptive feature selection module (AFSM) is proposed, to automatically learn the way to fuse multi-level representations in the channel dimension, in a data-driven manner, which significantly improves the performance of the detectors that have a feature pyramid structure, while introducing nearly free inference overhead.
UAV-Based Crowd Surveillance in Post COVID-19 Era
A complete UAV framework for intelligent monitoring of post COVID-19 outdoor activities is proposed and UAV trajectory design algorithm 2-Opt is recommended for practical real-time deployments due to its low-complexity and near-optimal performance.
Consistency-based Active Learning for Object Detection
Extensive experiments show that CALD outperforms existing state-of-the-art task-agnostic and detection-specific active learning methods on general object detection datasets.
ACE: Ally Complementary Experts for Solving Long-Tailed Recognition in One-Shot
A one-stage long-tailed recognition scheme, ally complementary experts (ACE), where the expert is the most knowledgeable specialist in a sub-set that dominates its training, and is complementary to other experts in the less-seen categories without being disturbed by what it has never seen is proposed.
The State of Aerial Surveillance: A Survey
This paper provides a comprehensive overview of human-centric aerial surveillance tasks from a computer vision and pattern recognition perspective and discusses unique challenges in performing these tasks in an aerial setting compared to a ground-based setting.
Deep Learning for Unmanned Aerial Vehicle-Based Object Detection and Tracking: A survey
This paper presents a meta-modelling framework that automates the very labor-intensive and therefore time-heavy and expensive process of manually cataloging and cataloging objects in UAVs.


Dense and Small Object Detection in UAV Vision Based on Cascade Network
A novel processing pipeline is proposed, that brings together several key contributions including the introduction of DeForm convolution layers within backbone; use of the interleaved cascade architecture; data augmentation process based on crop functionality and multi-model fusion of sub-category detection networks.
Patch-Level Augmentation for Object Detection in Aerial Images
A hard chip mining method is proposed that makes the ratio of each class balanced and generates hard examples that are efficient for model training and achieves superior results on VisDrone dataset both qualitatively and quantitatively.
Overcoming Classifier Imbalance for Long-Tail Object Detection With Balanced Group Softmax
  • Yu Li, Tao Wang, Jiashi Feng
  • Computer Science
    2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2020
This work provides the first systematic analysis on the underperformance of state-of-the-art models in front of long-tail distribution and proposes a novel balanced group softmax (BAGS) module for balancing the classifiers within the detection frameworks through group-wise training.
The Devil is in Classification: A Simple Framework for Long-tail Instance Segmentation
This work systematically investigates performance drop of the state-of-the-art two-stage instance segmentation model Mask R-CNN on the recent long-tail LVIS dataset, and unveils that a major cause is the inaccurate classification of object proposals.
Clustered Object Detection in Aerial Images
This paper proposes a Clustered Detection (ClusDet) network that unifies object clustering and detection in an end-to-end framework and achieves promising performance in comparison with state-of-the-art detectors.
The Unmanned Aerial Vehicle Benchmark: Object Detection and Tracking
This work constructs a new UAV benchmark focusing on complex scenarios with new level challenges, and is the first time to explore issues in unconstrained scenes comprehensively in UAV based real scenes.
Classification Calibration for Long-tail Instance Segmentation
This report investigates the performance drop phenomenon of state-of-the-art two-stage instance segmentation models when processing extreme long-tail training data based on the LVIS dataset, and finds a major cause is the inaccurate classification of object proposals.
Density Map Guided Object Detection in Aerial Images
This paper proposes a Density-Map guided object detection Network (DMNet), which is inspired from the observation that the object density map of an image presents how objects distribute in terms of the pixel intensity of the map.
Vision Meets Drones: Past, Present and Future
The VisDrone dataset, which is captured over various urban/suburban areas of 14 different cities across China from North to South, is described, being the largest such dataset ever published, and enables extensive evaluation and investigation of visual analysis algorithms on the drone platform.
The Power of Tiling for Small Object Detection
This study addresses the detection of pedestrians and vehicles onboard a micro aerial vehicle (MAV) with high-resolution imagery by exploiting PeleeNet, to best knowledge the most efficient network model on mobile GPUs, as the backbone of an SSD network as well as 38x38 feature map in the earlier layer.