• Journal of Information Processing Systems
• /
• v.12 no.4
• /
• pp.754-764
• /
• 2016

The exhaustive list of sparsification methods for a digital image suffers from achieving an adequate number of zero and near-zero coefficients. The method proposed in this paper, which is known as the Discrete Rajan Transform Sparsification, overcomes this inadequacy. An attempt has been made to compare the simulation results for benchmark images by various popular, existing techniques and analyzing from different aspects. With the help of Discrete Rajan Transform algorithm, both lossless and lossy sparse representations are obtained. We divided an image into $8{\times}8-sized$ blocks and applied the Discrete Rajan Transform algorithm to it to get a more sparsified spectrum. The image was reconstructed from the transformed output of the Discrete Rajan Transform algorithm with an acceptable peak signal-to-noise ratio. The performance of the Discrete Rajan Transform in providing sparsity was compared with the results provided by the Discrete Fourier Transform, Discrete Cosine Transform, and the Discrete Wavelet Transform by means of the Degree of Sparsity. The simulation results proved that the Discrete Rajan Transform provides better sparsification when compared to other methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.12
• /
• pp.3904-3922
• /
• 2022

As a research hotspot, pedestrian detection has a wide range of applications in the field of computer vision in recent years. However, current pedestrian detection methods have problems such as insufficient detection accuracy and large models that are not suitable for large-scale deployment. In view of these problems mentioned above, a lightweight pedestrian detection and early warning method using a new model called you only look once (Yolov5) is proposed in this paper, which utilizing advantages of Yolov5s model to achieve accurate and fast pedestrian recognition. In addition, this paper also optimizes the loss function of the batch normalization (BN) layer. After sparsification, pruning and fine-tuning, got a lot of optimization, the size of the model on the edge of the computing power is lower equipment can be deployed. Finally, from the experimental data presented in this paper, under the training of the road pedestrian dataset that we collected and processed independently, the Yolov5s model has certain advantages in terms of precision and other indicators compared with traditional single shot multiBox detector (SSD) model and fast region-convolutional neural network (Fast R-CNN) model. After pruning and lightweight, the size of training model is greatly reduced without a significant reduction in accuracy, and the final precision reaches 87%, while the model size is reduced to 7,723 KB.