Scale-aware Automatic Augmentation for Object Detection

Chen, Yukang; Li, Yanwei; Kong, Tao; Qi, Lu; Chu, Ruihang; Li, Lei; Jia, Jiaya

doi:10.1109/cvpr46437.2021.00944

Cited by 41 publications

(20 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This already suggests the importance of applying zoom-in and zoom-out data augmentation in the training phase. 63 Additionally, training on 100 × 100, 200 × 200 and 300 × 300 and 500 × 500 patches further improved the performance to ≈75%.…”

Section: Resultsmentioning

confidence: 97%

Convolutional neural networks for high throughput screening of catalyst layer inks for polymer electrolyte fuel cells

et al. 2021

View full text Add to dashboard Cite

show abstract

Section: Resultsmentioning

confidence: 97%

Convolutional neural networks for high throughput screening of catalyst layer inks for polymer electrolyte fuel cells

et al. 2021

View full text Add to dashboard Cite

show abstract

“…the recognition accuracy achieves 99.42% after training on the color feature augmented datasets. In [3], a universally applicable adaptive color feature augmentation algorithm is proposed for the target detection task. The model gets more representations belonging to the target in color field by leveraging the Gaussian filter in the target area, thus the color features of the targets is significantly expanded.…”

Section: Figure 7 Samples Of Random Contrastmentioning

confidence: 99%

“…DA (Data augmentation) algorithm, one of the most efficient methods, offers a sufficient proposal to solve such a problem. For example, in [2] and [3], an automatic augmentation algorithm based on reinforcement learning is adopted to improve the precision of object detection by 2-5% successfully; in [4], model based on improved GANs (Generation Adversarial Nets) is designed to execute a fault diagnose task, furthermore, it realizes update and auto prediction of various zero-shot fault modes. The challenge of insufficient data becomes more and more prominent, with the deep integration of CV technology and all walks of life gradually.…”

Section: Introductionmentioning

confidence: 99%

The Status and Challenges of Image Data Augmentation Algorithms

Jiang

Wang

Song

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Image data-augmentation algorithms effectively trump the problem of insufficient training samples for deep learning in some application fields, and it is typically for scholars to choose some of them for various computer vision tasks. But as the algorithms develop rapidly, the early proposed classification that the data-augmentation algorithms are sorted into classical ways and generating methods is no more suitable, because such classification misses some other meaningful strategies. Besides, it is frustrating for someone to decide which is the exact method to undertake, though there are too many optional algorithms to choose. Towards the goal of making some suggestions, the paper categorizes image data-augmentation algorithms into three kinds from the perspective of algorithm strategy, and they are matrix transformation algorithm, feature expansion algorithm and model generation based on neural network algorithm. The paper analyzes the typical algorithm principle, performance, application scenarios, research status and future challenges, and forecasts the development trend of data augmentation algorithms. The paper can provide academic reference for data augmentation algorithm in the fields of medicine and military.

show abstract

“…In complex environments, feature extraction of traffic signs is susceptible to various noise types, and the proportion of traffic signs in the whole image is very limited. erefore, multiscale feature extraction is particularly important in TSR [30]. FPNs are the basic component of the recognition system for detecting objects of different scales.…”

Section: Tsd Based On Cnnmentioning

confidence: 99%

Multiscale Traffic Sign Detection Method in Complex Environment Based on YOLOv4

Wang

Bai

Wang

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Traffic sign detection is a challenging problem in the field of unmanned driving, particularly important in complex environments. We propose a method, based on the improved You only look once (YOLO) v4, to detect and recognize multiscale traffic signs in complex environments. This method employs an image preprocessing module that can classify and denoize images of complex environments and then input the images into the improved YOLOv4. We also design an improved feature pyramid structure to replace the original feature pyramid of YOLOv4. This structure uses an adaptive feature fusion module and a multiscale feature transfer mechanism to reduce putative information loss in the feature map generation process and improve the information transfer between deep and shallow features, enhancing the representation ability of feature pyramids. Finally, we use EIOU LOSS and Cluster-NMS to further improve the model performance. The experimental results on the fusion of Tsinghua-Tencent 100 K and our collected dataset show that the proposed method achieves an mAP of 81.78%. Compared to existing methods, our method demonstrates its superiority with regard to traffic sign detection.

show abstract

Scale-aware Automatic Augmentation for Object Detection

Cited by 41 publications

References 26 publications

Convolutional neural networks for high throughput screening of catalyst layer inks for polymer electrolyte fuel cells

Convolutional neural networks for high throughput screening of catalyst layer inks for polymer electrolyte fuel cells

The Status and Challenges of Image Data Augmentation Algorithms

Multiscale Traffic Sign Detection Method in Complex Environment Based on YOLOv4

Contact Info

Product

Resources

About