yisun98 / Awesome-Multimodal-Fusion

Awesome-RGB-T-IR-Fusion

Multispectral-Panchromatic Fusion (Pan-sharpening)

RGB-Infrared Fusion

Dataset

1. DroneVieche
2. Vedia
3. KSITT

Method

Natural Scene (Perceptual-orientated)

1. Sample

2. Advantages

3. Disadvantages 追求融合图像视觉质量和评估指标的提升，而忽略了下游任务（如目标检测、目标分割等）的需求； [low-level tasks和high-level tasks之间存在gap]

Detection Scene (Camare-based, Task-orientated)

1. Sample

2. Advantages

3. Disadvantages 难以解决鸟瞰视角下的一些挑战； [迁移性弱]

Detection Scene (UAV-based, Task-orientated, More Complex Scene)

关键问题

1. 模态特征冗余度高
2. 模态差异大，融合特征难

Related Works

1. N. Zhang, Y. Liu, H. Liu, T. Tian, and J. Tian, “Oriented Infrared Vehicle Detection in Aerial Images via Mining Frequency and Semantic Information,” IEEE Transactions on Geoscience and Remote Sensing, vol. 61, no. 1, pp. 1-15, 2023.
2. Z. Huang, W. Li, and R. Tao, “Multimodal Knowledge Distillation for Arbitrary-Oriented Object Detection in Aerial Images,” in Proc. IEEE International Conference on Acoustics, Speech and Signal Processing, 2023, pp. 1-5.