Tag：paper
-
Time：2020-11-26
Jigsaw pre training generates data sets for backbone network pre training in a jigsaw way from the detection data set, without the need for additional pre training data sets, such as Imagenet. In addition, in order to make the network better adapt to the mosaic data, this paper proposes the ERF adaptive dense classification method, […]
-
Time：2020-11-24
introduction The dcgan network introduced in this paper is based on GaN network and introduces image convolution operation, that is, it combines CNN and Gan network well, so that the network generator can finally generate a false image from random noise. The author of the paperAlec Radford & Luke Metz。 In this paper, the author […]
-
Time：2020-11-22
By Nathan LambertCompile | VKSource: toward Data Science 1. Supervised learning Modern intensive learning is almost completely concentrated on deep reinforcement learning. The word “deep” in deep reinforcement learning means the use of neural networks at the core of the algorithm. In the process of learning, neural network makes some high-dimensional approximation. Having said that, […]
-
Time：2020-11-22
The work of this paper belongs to the application of deep learning in the industrial field. Referring to the solution of computer vision, the adaptive aprelu is proposed for the scene of machine fault detection, which greatly improves the accuracy of fault detection. The overall idea of the paper should also be applied to computer […]
-
Time：2020-11-19
In this paper, a dynamic relu is proposed, which can dynamically adjust the corresponding segmented activation function according to the input. Compared with relu and its variants, only a small amount of extra calculation can bring about significant performance improvement, and can be embedded into the current mainstream model seamlessly Source: Xiaofei’s algorithm Engineering Notes […]
-
Time：2020-11-19
In order to reduce the computational cost of neural network, ghost module is proposed to construct efficient network results. In this module, the original convolution layer is divided into two parts. Firstly, fewer convolution kernels are used to generate a small number of intrinsic feature maps, and then a simple linear change operation is used […]
-
Time：2020-11-16
introduction This paper introduces a U-shaped network structure for semantic segmentation. In fact, it is based on the idea of encoding and decoding, which can effectively combine the information of low resolution and high resolution, and can better segment the image edge. It was proposed in the same year as FCN, but u-net uses a […]
-
Time：2020-11-15
In this paper, DBTD method is used to calculate the filtering threshold, and then the random pruning algorithm is combined to prune the eigenvalue gradient, which can reduce the calculation amount in the backhaul phase. The training on CPU and arm has 3.99 times and 5.92 times acceleration effect respectively Source: Xiaofei’s algorithm Engineering Notes […]
-
Time：2020-11-11
This paper proposes a new target detection algorithm with less samples. The innovations include attention RPN, multi relation detector and comparative training strategy. In addition, we construct a small sample detection dataset fsod containing 1000 classes. The paper model trained on fsod can be directly transferred to the detection of new categories without fine tune […]
-
Time：2020-11-10
Although the new coronavirus is not as powerful as SARS in 2002, its terrible feature is that it has been latent for a long time without symptoms, which makes the infected people spread the virus without knowing it. If there is no strong means of prevention, the spread of the virus is almost impossible to […]
-
Time：2020-11-10
In this paper, the Piou loss function extended from IOU index is proposed, which can effectively improve the rotation angle prediction and IOU effect in the scene of tilt target detection. It is suitable for anchor based method and anchor free method. In addition, the retail50k dataset is provided, which can be used to evaluate […]
-
Time：2020-11-3
In this paper, we improve the real-time branch rate of yolcat and yoltat, but it is worth to improve the real-time performance of yoltat and yoltat in real-time Source: Xiaofei’s algorithm Engineering Notes official account Thesis: yoact + + better real-time instance segmentation Thesis address: https://arxiv.org/pdf/1912.06218.pdf **Paper code: https://github.com/dbolya/yolact** Introduction Although the former yolact can […]
