Material recognition is researched in both computervision and vision science fields. In this paper, we investigated how humans observe material images and found the eye fixation information improves the performance o...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
Material recognition is researched in both computervision and vision science fields. In this paper, we investigated how humans observe material images and found the eye fixation information improves the performance of material image classification models. We first collected eye-tracking data from human observers and used it to fine-tune a generative adversarial network for saliency prediction (SalGAN). We then fused the predicted saliency map with material images and fed them to CNN models for material classification. The experiment results show that the classification accuracy is improved than those using original images. This indicates that human's visual cues could benefit computational models as priors.
The land cover classification task of the DeepGlohe Challenge presents significant obstacles even to state of the art segmentation models due to a small amount of data, incomplete and sometimes incorrect labeling, and...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
The land cover classification task of the DeepGlohe Challenge presents significant obstacles even to state of the art segmentation models due to a small amount of data, incomplete and sometimes incorrect labeling, and highly imbalanced classes. In this work, we show an approach based on the U-Net architecture with the Lovcisz-Softmax loss that successfully alleviates these problems: we compare several different convolutional architectures for U-Net encoders.
In this paper, we study deep transfer learning as a way of overcoming object recognition challenges encountered in the field of digital pathology. Through several experiments, we investigate various uses of pre-traine...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
In this paper, we study deep transfer learning as a way of overcoming object recognition challenges encountered in the field of digital pathology. Through several experiments, we investigate various uses of pre-trained neural network architectures and different combination schemes with random forests for feature selection. Our experiments on eight classification datasets show that densely connected and residual networks consistently yield best performances across strategies. It also appears that network fine-tuning and using inner layers features are the best performing strategies, with the former yielding slightly superior results.
Deciphering human behaviors to predict their future paths/trajectories and what they would do from videos is important in many applications. With the advancement in deep learning, systems now are able to analyze an un...
详细信息
ISBN:
(数字)9781728125060
ISBN:
(纸本)9781728125077
Deciphering human behaviors to predict their future paths/trajectories and what they would do from videos is important in many applications. With the advancement in deep learning, systems now are able to analyze an unprecedented amount of rich visual information from videos. An important analysis is forecasting the future path of pedestrians, called future person trajectory prediction. This problem has received increasing attention in the computervision community. It is regarded as an essential building block in video understanding for many applications like self-driving cars, socially-aware robots, etc.
We present a semantic segmentation algorithm for RGB remote sensing images. Our method is based on the Dilated Stacked U-Nets architecture. This state-of-the-art method has been shown to have good performance in other...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
We present a semantic segmentation algorithm for RGB remote sensing images. Our method is based on the Dilated Stacked U-Nets architecture. This state-of-the-art method has been shown to have good performance in other applications. We perform additional post-processing by blending image tiles and degridding the result. Our method gives competitive results on the DeepGlobe dataset.
We introduce the first benchmark for a new problem - recognizing human action adverbs (HAA): "Adverbs Describing Human Actions" (ADHA). We demonstrate some key features of ADHA: a semantically complete set o...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
We introduce the first benchmark for a new problem - recognizing human action adverbs (HAA): "Adverbs Describing Human Actions" (ADHA). We demonstrate some key features of ADHA: a semantically complete set of adverbs describing human actions, a set of common, describable human actions, and an exhaustive labelling of simultaneously emerging actions in each video. We commit an in-depth analysis on the implementation of current effective models in action recognition and image captioning on adverb recognition, and the results reveal that such methods are unsatisfactory. Furthermore, we propose a novel three-stream hybrid model to tackle the HAA problem, which achieves better performances and receives relatively promising results.
We propose ViDeNN: a CNN for Video Denoising without prior knowledge on the noise distribution (blind denoising). The CNN architecture uses a combination of spatial and temporal filtering, learning to spatially denois...
详细信息
ISBN:
(数字)9781728125060
ISBN:
(纸本)9781728125077
We propose ViDeNN: a CNN for Video Denoising without prior knowledge on the noise distribution (blind denoising). The CNN architecture uses a combination of spatial and temporal filtering, learning to spatially denoise the frames first and at the same time how to combine their temporal information, handling objects motion, brightness changes, low-light conditions and temporal inconsistencies. We demonstrate the importance of the data used for CNNs training, creating for this purpose a specific dataset for low-light conditions. We test ViDeNN on common benchmarks and on self-collected data, achieving good results comparable with the state-of-the-art.
Millions of people are disconnected from basic services due to lack of adequate addressing. We propose an automatic generative algorithm to create street addresses from satellite imagery. Our addressing scheme is cohe...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
Millions of people are disconnected from basic services due to lack of adequate addressing. We propose an automatic generative algorithm to create street addresses from satellite imagery. Our addressing scheme is coherent with the street topology, linear and hierarchical to follow human perception, and universal to be used as a unified geocoding system. Our algorithm starts with extracting road segments using deep learning and partitions the road network into regions. Then regions, streets, and address cells are named using proximity computations. We also extend our addressing scheme to cover inaccessible areas, to be flexible for changes, and to lead as a pioneer for a unified geodatabase.
Cross domain image retrieval is a challenging task that implies matching images from one domain to their pairs from another domain. In this paper we focus on fashion image retrieval, which involves matching an image o...
详细信息
ISBN:
(数字)9781538661000
ISBN:
(纸本)9781538661000
Cross domain image retrieval is a challenging task that implies matching images from one domain to their pairs from another domain. In this paper we focus on fashion image retrieval, which involves matching an image of a fashion item taken by users, to the images of the same item taken in controlled condition, usually by professional photographer. When facing this problem, we have different products in train and test time, and we use triplet loss to train the network. We stress the importance of proper training of simple architecture, as well as adapting general models to the specific task.
暂无评论