DSConv: Efficient Convolution Operator

Nascimento, Marcelo Gennari do; Prisacariu, Victor Adrian; Fawcett, Roger

doi:10.1109/iccv.2019.00525

Cited by 63 publications

(24 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Where Conv-Block is the standard convolution module, DSconv is the depthwise separable convolution [13], Up-Conv is the up-convolution, and Down-Conv is the down-convolution. There are Residual blocks between the encoder and decoder linear bottleneck residual blocks.…”

Section: Architecture Of Animationganmentioning

confidence: 99%

Linear-ResNet GAN-based anime style transfer of face images

Chen

Dai

Wei

et al. 2023

SIViP

View full text Add to dashboard Cite

Converting directly real-world images into high-quality anime styles using generative adversarial networks is one of the research hotspots in computer vision. The current popular AnimeGAN and WhiteBox anime generative adversarial networks have distortion of image features problem and loss of details on lines and textures problem, respectively. To address these problems, we introduce a new AnimationGAN based on a linear bottleneck residual network and a hybrid attention mechanism. The proposed AnimationGAN can prevent severe detail loss and distortion of image features in AnimeGAN-transferred images. In addition, we adopt optimized normalizations to improve the accuracy and learning rate of the model. The experimental results show that compared with AnimeGAN and Whitebox, the generated animation image significantly improved line texture details and image feature retention, and the network training speed is also faster.

show abstract

Section: Architecture Of Animationganmentioning

confidence: 99%

Linear-ResNet GAN-based anime style transfer of face images

Chen

Dai

Wei

et al. 2023

SIViP

View full text Add to dashboard Cite

show abstract

“…This strategy has been applied to CNNs and tensors before. Previous papers have used different strategies for sharing exponents: for inference only [5,27,28], and for training as well [4,6,32,42]. Most of the methods share the exponent across entire tensors, or batches.…”

Section: Preamblementioning

confidence: 99%

“…We believe that these options are unnecessarily restrictive, and more accuracy can be obtained by using smaller blocks. We follow the strategy of [27], in which exponents are shared channel-wise. This allows for a fine-grained trade-off between accuracy and computation, whilst taking advantage of fixed point operations in hardware.…”

Section: Preamblementioning

confidence: 99%

HyperBlock Floating Point: Generalised Quantization Scheme for Gradient and Inference Computation

Nascimento

Prisacariu

Fawcett

et al. 2023

2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

Self Cite

View full text Add to dashboard Cite

Prior quantization methods focus on producing networks for fast and lightweight inference. However, the cost of unquantised training is overlooked, despite requiring significantly more time and energy than inference. We present a method for quantizing convolutional neural networks for efficient training. Quantizing gradients is challenging because it requires higher granularity and their values span a wider range than the weight and feature maps. We propose an extension of the Channel-wise Block Floating Point format that allows for quick gradient computation, using a minimal amount of quantization time. This is achieved through sharing an exponent across both depth and batch dimensions in order to quantize tensors once and reuse them during backpropagation. We test our method using standard models such as AlexNet, VGG, and ResNet, on the CI-FAR10, SVHN and ImageNet datasets. We show no loss of accuracy when quantizing AlexNet weights, activations and gradients to only 4 bits training ImageNet.

show abstract

“…In CAM modules, the receptive field is expanded, and the computation also increases. To alleviate this problem, we introduce the highly efficient convolution operator DSConv proposed in reference [43]. DSConv is a more efficient quantization operator that replaces single precision operations with integer operations, whereas it is able to preserve both the kernel weights and the probability distribution.…”

Section: Dsconvmentioning

confidence: 99%

ACD‐YOLO: Improved YOLOv5‐based method for steel surface defects detection

Fan,

Wang,

et al. 2023

IET Image Processing

View full text Add to dashboard Cite

Since the quality of steel is of paramount importance in modern production, the defects detection of steel surface is significantly crucial. In this field, two‐stage detection algorithms have encountered issues about low detection speed, while one‐stage detection algorithms have room for improvement in detection accuracy. How to trade‐off between accuracy and speed of detection to better meet the demands of industrial production remains a challenge. To address this problem, this paper proposes a You Only Look Once version 5 (YOLOv5)‐based improved method ACD‐YOLO. ACD‐YOLO model incorporates anchors optimization, context augmentation module, and efficient convolution operators. In anchor optimization, the boundaries of anchors are optimized using an improved genetic algorithm. Moreover, to improve detection accuracy, a context augmentation module is incorporated into both the head and the backbone end of the network. Additionally, efficient convolution operators are adopted to address the increase of computation complexity caused by the context augmentation module. Experimental results show that ACD‐YOLO achieves mean average precision of 79.3%, with frames per second of 72. Compared to reference methods, ACD‐YOLO achieves the best balance between accuracy and speed of detection, and is more suitable for practice industrial production.

show abstract

DSConv: Efficient Convolution Operator

Cited by 63 publications

References 13 publications

Linear-ResNet GAN-based anime style transfer of face images

Linear-ResNet GAN-based anime style transfer of face images

HyperBlock Floating Point: Generalised Quantization Scheme for Gradient and Inference Computation

ACD‐YOLO: Improved YOLOv5‐based method for steel surface defects detection

Contact Info

Product

Resources

About