Partially Fake it Till you Make It

Bongini, Francesco; Berlincioni, Lorenzo; Bertini, Marco; Bimbo, Alberto Del

doi:10.1145/3474085.3475679

Cited by 11 publications

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To verify the effectiveness of our method, we compared it with several existing state-of-the-art methods, as described in this subsection, using the FLIR dataset for all experimental results. These methods mainly include image fusion methods [ 35 , 36 , 37 , 38 , 39 ], image enhancement [ 23 ], and image generation [ 8 ]. From Table 1 , we can draw several conclusions: (1) The performance of the two detection networks based on image-enhancement or image-generation methods is slightly inferior to that of the image fusion method.…”

Section: Methodsmentioning

confidence: 99%

“…Such a technique can dramatically increase detection accuracy while speeding up network convergence on a limited dataset. In order to increase the total number of infrared data in the target domain and enhance the precision of infrared object detection, some researchers use image-generation techniques to transform visible image data into pseudo-infrared images [ 8 ]. In addition, image fusion methods are used by researchers to fuse visible and infrared modalities in the network [ 9 , 10 , 11 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Task-Decoupled Knowledge Transfer for Cross-Modality Object Detection

Wei

Bai

Chen

et al. 2023

Entropy

View full text Add to dashboard Cite

In harsh weather conditions, the infrared modality can supplement or even replace the visible modality. However, the lack of a large-scale dataset for infrared features hinders the generation of a robust pre-training model. Most existing infrared object-detection algorithms rely on pre-training models from the visible modality, which can accelerate network convergence but also limit performance due to modality differences. In order to provide more reliable feature representation for cross-modality object detection and enhance its performance, this paper investigates the impact of various task-relevant features on cross-modality object detection and proposes a knowledge transfer algorithm based on classification and localization decoupling analysis. A task-decoupled pre-training method is introduced to adjust the attributes of various tasks learned by the pre-training model. For the training phase, a task-relevant hyperparameter evolution method is proposed to increase the network’s adaptability to attribute changes in pre-training weights. Our proposed method improves the accuracy of multiple modalities in multiple datasets, with experimental results on the FLIR ADAS dataset reaching a state-of-the-art level and surpassing most multi-spectral object-detection methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Task-Decoupled Knowledge Transfer for Cross-Modality Object Detection

Wei

Bai

Chen

et al. 2023

Entropy

View full text Add to dashboard Cite

show abstract

“…For instance, Kieu et al [25] proposed to synthesize realistic thermal versions of input RGB images through a Generative Adversarial Network (GAN) and then mixed real and improved fake thermal images as a way of data augmentation to relieve the problem of limited thermal image dataset in object detection. Similarly, Bongini et al [35] also proposed to produce synthetic thermal data by rendering 3D models using a thermal shader in the Unity game engine, and then utilized GAN to improve fake thermal image realism. TC Det [16] established an auxiliary branch of day-and-night prediction to guide the domain adaptation from visible to thermal.…”

Section: A Thermal-based Pedestrian Detectionmentioning

confidence: 99%

Illumination-Aware Hallucination-Based Domain Adaptation for Thermal Pedestrian Detection

Xie,

Cheng,

Dai

et al. 2024

IEEE Trans. Intell. Transport. Syst.

View full text Add to dashboard Cite

Thermal imagery is emerging as a viable candidate for 24-7, all-weather pedestrian detection owning to thermal sensors' robust performance for pedestrian detection under different weather and illumination conditions. Despite the promising results obtained from combining visible (RGB) and thermal cameras in multi-spectral fusion techniques, the complex synchronization requirements, including alignment and calibration of sensors, impede their deployment in real-world scenarios. In this paper, we introduce a novel approach for domain adaptation to enhance the performance of pedestrian detection based solely on thermal images. Our proposed approach involves several stages. Firstly, we use both thermal and visible images as input during the training phase. Secondly, we leverage a thermal-to-visible hallucination network to generate feature maps that are similar to those generated by the visible branch. Finally, we design a transformer-based multi-modal fusion module to integrate the hallucinated visible and thermal information more effectively. The thermal-to-visible hallucination network acts as domain adaptation, allowing us to obtain pseudo-visual and thermal features using solely thermal input. Based on the experimental results, it is observed the mean average precision (mAP) increases by 4.72% and the miss rate decreases by 7.56% on the KAIST dataset when compared to the baseline model.

show abstract

A Survey of Synthetic Data Augmentation Methods in Machine Vision

Mumuni,

Gerrar

2024

Mach. Intell. Res.

View full text Add to dashboard Cite

Partially Fake it Till you Make It

Cited by 11 publications

References 45 publications

Task-Decoupled Knowledge Transfer for Cross-Modality Object Detection

Task-Decoupled Knowledge Transfer for Cross-Modality Object Detection

Illumination-Aware Hallucination-Based Domain Adaptation for Thermal Pedestrian Detection

A Survey of Synthetic Data Augmentation Methods in Machine Vision

Contact Info

Product

Resources

About