Wei-ping Zhu

Bo Shi, Wei-ping Zhu, M. N. S. Swamy

Spatially variant dynamic convolution provides a principled approach of integrating spatial adaptivity into deep neural networks. However, mainstream designs in medical segmentation commonly generate dynamic kernels through average pooling, which implicitly collapses high-frequency spatial details into a coarse, spatially-compressed representation, leading to over-smoothed predictions that degrade the fidelity of fine-grained clinical structures. To address this limitation, we propose a novel Structure-Guided Dynamic Convolution (SGDC) mechanism, which leverages an explicitly supervised structure-extraction branch to guide the generation of dynamic kernels and gating signals for structure-aware feature modulation. Specifically, the high-fidelity boundary information from this auxiliary branch is fused with semantic features to enable spatially-precise feature modulation. By replacing context aggregation with pixel-wise structural guidance, the proposed design effectively prevents the information loss introduced by average pooling. Experimental results show that SGDC achieves state-of-the-art performance on ISIC 2016, PH2, ISIC 2018, and CoNIC datasets, delivering superior boundary fidelity by reducing the Hausdorff Distance (HD95) by 2.05, and providing consistent IoU gains of 0.99\%-1.49\% over pooling-based baselines. Moreover, the mechanism exhibits strong potential for extension to other fine-grained, structure-sensitive vision tasks, such as small-object detection, offering a principled solution for preserving structural integrity in medical image analysis. To facilitate reproducibility and encourage further research, the implementation code for both our SGE and SGDC modules has been is publicly released at https://github.com/solstice0621/SGDC.

Md. Rabiul Islam, Daniel Massicotte, Philippe Y. Massicotte et al.

Gesture recognition using low-resolution instantaneous HD-sEMG images opens up new avenues for the development of more fluid and natural muscle-computer interfaces. However, the data variability between inter-session and inter-subject scenarios presents a great challenge. The existing approaches employed very large and complex deep ConvNet or 2SRNN-based domain adaptation methods to approximate the distribution shift caused by these inter-session and inter-subject data variability. Hence, these methods also require learning over millions of training parameters and a large pre-trained and target domain dataset in both the pre-training and adaptation stages. As a result, it makes high-end resource-bounded and computationally very expensive for deployment in real-time applications. To overcome this problem, we propose a lightweight All-ConvNet+TL model that leverages lightweight All-ConvNet and transfer learning (TL) for the enhancement of inter-session and inter-subject gesture recognition performance. The All-ConvNet+TL model consists solely of convolutional layers, a simple yet efficient framework for learning invariant and discriminative representations to address the distribution shifts caused by inter-session and inter-subject data variability. Experiments on four datasets demonstrate that our proposed methods outperform the most complex existing approaches by a large margin and achieve state-of-the-art results on inter-session and inter-subject scenarios and perform on par or competitively on intra-session gesture recognition. These performance gaps increase even more when a tiny amount (e.g., a single trial) of data is available on the target domain for adaptation. These outstanding experimental results provide evidence that the current state-of-the-art models may be overparameterized for sEMG-based inter-session and inter-subject gesture recognition tasks.

Kai Wang, Bengbeng He, Wei-Ping Zhu

In this paper, we propose a transformer-based architecture, called two-stage transformer neural network (TSTNN) for end-to-end speech denoising in the time domain. The proposed model is composed of an encoder, a two-stage transformer module (TSTM), a masking module and a decoder. The encoder maps input noisy speech into feature representation. The TSTM exploits four stacked two-stage transformer blocks to efficiently extract local and global information from the encoder output stage by stage. The masking module creates a mask which will be multiplied with the encoder output. Finally, the decoder uses the masked encoder feature to reconstruct the enhanced speech. Experimental results on the benchmark dataset show that the TSTNN outperforms most state-of-the-art models in time or frequency domain while having significantly lower model complexity.

Farnood Faraji, Yazid Attabi, Benoit Champagne et al.

The advent of learning-based methods in speech enhancement has revived the need for robust and reliable training features that can compactly represent speech signals while preserving their vital information. Time-frequency domain features, such as the Short-Term Fourier Transform (STFT) and Mel-Frequency Cepstral Coefficients (MFCC), are preferred in many approaches. While the MFCC provide for a compact representation, they ignore the dynamics and distribution of energy in each mel-scale subband. In this work, a speech enhancement system based on Generative Adversarial Network (GAN) is implemented and tested with a combination of Audio FingerPrinting (AFP) features obtained from the MFCC and the Normalized Spectral Subband Centroids (NSSC). The NSSC capture the locations of speech formants and complement the MFCC in a crucial way. In experiments with diverse speakers and noise types, GAN-based speech enhancement with the proposed AFP feature combination achieves the best objective performance while reducing memory requirements and training time.

Md. Rabiul Islam, Daniel Massicotte, Francois Nougarou et al.

The concept of neuromuscular activity recognition using instantaneous high-density surface electromyography (HD-sEMG) images opens up new avenues for the development of more fluid and natural muscle-computer interfaces. However, the existing approaches employed a very large deep convolutional neural network (ConvNet) architecture and complex training schemes for HD-sEMG image recognition, which requires the network architecture to be pre-trained on a very large-scale labeled training dataset, as a result, it makes computationally very expensive. To overcome this problem, we propose S-ConvNet and All-ConvNet models, a simple yet efficient framework for learning instantaneous HD-sEMG images from scratch for neuromuscular activity recognition. Without using any pre-trained models, our proposed S-ConvNet and All-ConvNet demonstrate very competitive recognition accuracy to the more complex state of the art for neuromuscular activity recognition based on instantaneous HD-sEMG images, while using a ~ 12 x smaller dataset and reducing learning parameters to a large extent. The experimental results proved that the S-ConvNet and All-ConvNet are highly effective for learning discriminative features for instantaneous HD-sEMG image recognition especially in the data and high-end resource constrained scenarios.

Md Tauhidul Islam, Asaduzzaman, Celia Shahnaz et al.

A two-step enhancement method based on spectral subtraction and phase spectrum compensation is presented in this paper for noisy speeches in adverse environments involving non-stationary noise and medium to low levels of SNR. The magnitude of the noisy speech spectrum is modified in the first step of the proposed method by a spectral subtraction approach, where a new noise estimation method based on the low frequency information of the noisy speech is introduced. We argue that this method of noise estimation is capable of estimating the non-stationary noise accurately. The phase spectrum of the noisy speech is modified in the second step consisting of phase spectrum compensation, where an SNR-dependent approach is incorporated to determine the amount of compensation to be imposed on the phase spectrum. A modified complex spectrum is obtained by aggregating the magnitude from the spectral subtraction step and modified phase spectrum from the phase compensation step, which is found to be a better representation of enhanced speech spectrum. Speech files available in the NOIZEUS database are used to carry extensive simulations for evaluation of the proposed method.

Md Tauhidul Islam, Celia Shahnaz, Wei-Ping Zhu et al.

A speech enhancement method based on probabilistic geometric approach to spectral subtraction (PGA) performed on short time magnitude spectrum is presented in this paper. A confidence parameter of noise estimation is introduced in the gain function of the proposed method to prevent subtraction of the overestimated and underestimated noise, which not only removes the noise efficiently but also prevents the speech distortion. The noise compensated magnitude spectrum is then recombined with the unchanged phase spectrum to produce a modified complex spectrum prior to synthesize an enhanced frame. Extensive simulations are carried out using the speech files available in the NOIZEUS database in order to evaluate the performance of the proposed method.

Md Tauhidul Islam, Celia Shahnaz, Wei-Ping Zhu et al.

In this paper, for real time enhancement of noisy speech, a method of threshold determination based on modeling of Teager energy (TE) operated perceptual wavelet packet (PWP) coefficients of the noisy speech and noise by an Erlang-2 PDF is presented. The proposed method is computationally much faster than the existing wavelet packet based thresholding methods. A custom thresholding function based on a combination of mu-law and semisoft thresholding functions is designed and exploited to apply the statistically derived threshold upon the PWP coefficients. The proposed custom thresholding function works as a mu-law or a semisoft thresholding function or their combination based on the probability of speech presence and absence in a subband of the PWP transformed noisy speech. By using the speech files available in NOIZEUS database, a number of simulations are performed to evaluate the performance of the proposed method for speech signals in the presence of Gaussian white and street noises. The proposed method outperforms some of the state-of-the-art speech enhancement methods both at high and low levels of SNRs in terms of standard objective measures and subjective evaluations including formal listening tests.

Md Tauhidul Islam, Celia Shahnaz, Wei-Ping Zhu et al.

A divide and conquer strategy for enhancement of noisy speeches in adverse environments involving lower levels of SNR is presented in this paper, where the total system of speech enhancement is divided into two separate steps. The first step is based on noise compensation on short time magnitude and the second step is based on phase compensation. The magnitude spectrum is compensated based on a modified spectral subtraction method where the cross-terms containing spectra of noise and clean speech are taken into consideration, which are neglected in the traditional spectral subtraction methods. By employing the modified magnitude and unchanged phase, a procedure is formulated to compensate the overestimation or underestimation of noise by phase compensation method based on the probability of speech presence. A modified complex spectrum based on these two steps are obtained to synthesize a musical noise free enhanced speech. Extensive simulations are carried out using the speech files available in the NOIZEUS database in order to evaluate the performance of the proposed method. It is shown in terms of the objective measures, spectrogram analysis and formal subjective listening tests that the proposed method consistently outperforms some of the state-of-the-art methods of speech enhancement for noisy speech corrupted by street or babble noise at very low as well as medium levels of SNR.