Quantization, which involves bit-width reduction, is considered as one of the most effective approaches to rapidly and energy-efficiently deploy deep convolutional neural networks (DCNNs) on resource-constrained embedded hardware. However, bit-width reduction on the weights and activations of DCNNs seriously degrades accuracy. To solve this problem, in this paper we propose a mixed hardware-friendly quantization (MXQN) method that applies fixed-point quantization and logarithmic quantization for DCNNs without the necessity to retrain and fine-tune the DCNN. Our MXQN algorithm is a multi-staged process where, first, we employ a signal-to-quantization-noise ratio (SQNR) process as the metric to estimate the interplay between the parameter quantization errors of each layer and the overall model prediction accuracy. Then, we utilize a fixed-point quantization process to quantize weights, and depending on the SQNR metric we empirically select either a logarithmic or a fixed-point quantization process to quantize activations. For improved accuracy, we propose an optimized logarithmic quantization scheme that affords a fine-grained step size. We evaluate the performance of MXQN utilizing the VGG16 network on the MNIST, CIFAR-10, CIFAR-100, and the ImageNet datasets, as well as VGG19 and ResNet (ResNet18, ResNet34, ResNet50) networks on the ImageNet, and demonstrate that the MXQN-quantized DCNN despite not being retrained and fine-tuned, it still achieves high accuracy close to the original DCNN.

MXQN:Mixed quantization for reducing bit-width of weights and activations in deep convolutional neural networks

Dictionary learning has become a research hotspot. How to construct a robust dictionary is a key issue. In face recognition problem, differences in expressions, postures, and lighting conditions are key factors that affect the accuracy. Therefore, images of the same face can be very different in different situations. In real-world scenario, the samples of each face are very limited, which make it hard for the network to generalize well. Therefore, To solve the problem mentioned above, this paper proposes a method to construct a robust dictionary. In the method, virtual samples are generated to appropriately reflect the diversity of the face images, and based on this, two dictionaries are constructed and a corresponding fusion classification scheme is designed. The main advantages of this method are as follows: firstly, the simultaneous use of virtual samples and original samples can better reflect the facial appearance of each morphology, and the dictionaries obtained help to improve the robustness of the dictionary learning algorithm. Secondly, the proposed fusion classification scheme can give full play to the performance of the double dictionary learning algorithm. The results of out experiments show that the proposed algorithm is superior to some existing algorithms.

Dictionary learning and face recognition based on sample expansion

Object detection based on remote sensing imagery has become increasingly popular over the past few years. Unlike natural images taken by humans or surveillance cameras, the scale of remote sensing images is large, which requires the training and inference procedure to be on a cutting image. However, objects appearing in remote sensing imagery are often sparsely distributed and the labels for each class are imbalanced. This results in unstable training and inference. In this paper, we analyze the training characteristics of the remote sensing images and propose the fusion of the aggregated-mosaic training method, with the assigned-stitch augmentation and auto-target-duplication. In particular, based on the ground truth and mosaic image size, the assigned-stitch augmentation enhances each training sample with an appropriate account of objects, facilitating the smooth training procedure. Hard to detect objects, or those in classes with rare samples, are randomly selected and duplicated by the auto-target-duplication, which solves the sample imbalance or classes with insufficient results. Thus, the training process is able to focus on weak classes. We employ VEDAI and NWPU VHR-10, remote sensing datasets with sparse objects, to verify the proposed method. The YOLOv5 adopts the Mosaic as the augmentation method and is one of state-of-the-art detectors, so we choose Mosaic (YOLOv5) as the baseline. Results demonstrate that our method outperforms Mosaic (YOLOv5) by 2.72% and 5.44% on 512 × 512 and 1024 × 1024 resolution imagery, respectively. Moreover, the proposed method outperforms Mosaic (YOLOv5) by 5.48% under the NWPU VHR-10 dataset.

An Improved Aggregated-Mosaic Method for the Sparse Object Detection of Remote Sensing Imagery

Improved image representation and sparse representation for face recognition

This paper presents a sign language recognition system based on color stream and skeleton points. Several approaches have been established to address sign language recognition problems. However, most of the previous approaches still have poor recognition accuracy. The proposed approach uses Kinect sensor based on color stream and skeleton points from the depth stream to improved recognition accuracy. Techniques within this approach use hand trajectories and hand shapes in combating sign recognition challenges. Therefore, for a particular sign a representative feature vector is extracted, which consists of hand trajectories and hand shapes. A sparse dictionary learning algorithm, Label Consistent K-SVD (LC-KSVD) is applied to obtain a discriminative dictionary. Based on that, the system was further developed to a new classification approach for better results. The proposed system was fairly evaluated based on 21 sign words including one-handed signs and two-handed signs. It was observed that the proposed system gets high recognition accuracy of 98.25%, and obtained an average accuracy of 95.34% for signer independent recognition.
Keywords: Sign language, Color stream, Skeleton points, Kinect sensor, Discriminative dictionary.

/pdf/sign-language-recognition-using-kinect-sensor-based-on-color-w176maykrv.pdf

Sign language recognition using Kinect sensor based on color stream and skeleton points

It seems that for multiple available images of the same object, the pixel values at the same image position are almost always different, which is especially obvious for the deformable object. This implies that it will be not easy to correctly classify the deformable object. In order to extract salient features of images and improve the performance of image classification, a novel image classification algorithm is proposed in this paper. The algorithm can effectively preserve the large-scale information and global features of the original image, reduce the difference in different images of the same object, and significantly improve the accuracy of image classification. Firstly, the virtual image is generated by the new image representation procedure. Secondly, the image classification algorithm is used to obtain the corresponding classification scores of the original image and the virtual image, respectively. Finally, the ultimate classification score is obtained by a simple and efficient score fusion scheme. A large number of experiments on three widely used image databases show that the proposed algorithm outperforms other state-of-the-art algorithms in classification accuracy. At the same time, the algorithm has the advantages of simple implementation and high computational efficiency.

Improved image representation and sparse representation for image classification

In recent years, dictionary learning has received more and more attention in the study of face recognition. However, most dictionary learning algorithms directly use the original training samples to learn the dictionary, ignoring noise existing in the training samples. For example, there are differences between different images of the same subject due to changes in illumination, expression, etc. To address the above problems, this paper proposes the dictionary relearning algorithm (DRLA) based on locality constraint and label embedding, which can effectively reduce the influence of noise on the dictionary learning algorithm. In our proposed dictionary learning algorithm, first, the initial dictionary and coding coefficient matrix are directly obtained from the training samples, and then the original training samples are reconstructed by the product of the initial dictionary and coding coefficient matrix. Finally, the dictionary learning algorithm is reapplied to obtain a new dictionary and coding coefficient matrix, and the newly obtained dictionary and coding coefficient matrix are used for subsequent image classification. The dictionary reconstruction method can partially eliminate noise in the original training samples. Therefore, the proposed algorithm can obtain more robust classification results. The experimental results demonstrate that the proposed algorithm performs better in recognition accuracy than some state-of-the-art algorithms.

/pdf/a-dictionary-learning-algorithm-based-on-dictionary-1ghdxr9g1i.pdf

A Dictionary Learning Algorithm Based on Dictionary Reconstruction and Its Application in Face Recognition

In recent years, dictionary learning method has been widely applied to face recognition and achieved good performance. However, most dictionary learning methods have two problems. First, they focused on only the resolution of the original images and did not consider impact of different resolutions on dictionary performance. When the attained dictionary is used to solve practical application problems, the recognition result of real-world images that may have a large difference in resolution with the original images will be very disappointed. Second, function of the dictionary will decrease due to insufficient training samples. Considering the above problems, this paper proposes a multi-resolution dictionary learning method based on sample expansion. We convert the original images to different resolutions and generate a dictionary for each resolution. Similarly, a dictionary is also produced for each resolution of reasonable virtual images generated by the original images. Then, for a test sample, a simple and efficient score fusion scheme is used to combine scores of the original image and the virtual image to obtain the ultimate classification score. We have performed experiments on multiple face databases, and the results show that our method has better performance than some state-of-the-art methods.

Shijun Zheng

Papers

Improved image representation and sparse representation for image classification

A Dictionary Learning Algorithm Based on Dictionary Reconstruction and Its Application in Face Recognition

Multi-resolution dictionary learning method based on sample expansion and its application in face recognition