arXiv subjects

Fengjie Zhu

Publications and source records attributed to Fengjie Zhu.

12 recordsLinked to original sources

arXiv2026

Transsion's Speaker-Attributed Multilingual ASR System for the MLC-SLM 2026 Challenge

This paper presents the Transsion Speech Team submission to Task 1 of the MLC-SLM 2026 Challenge, which focuses on speaker-attributed transcription for multilingual conversational speech. We propose a cascaded framework consisting of three components: a speaker diarization module, a long-form multilingual ASR module, and a speaker-transcription fusion module. The diarization module is built upon DiariZen and produces speaker-homogeneous segments through local speaker activity estimation and global speaker clustering. The ASR module is based on Qwen3-Omni and generates multilingual transcriptions, while an external CTC-based alignment model provides precise word- and character-level timestamps. Finally, the fusion module combines diarization outputs with timestamped transcriptions to generate speaker-attributed STM outputs. Experimental results on the official evaluation set demonstrate the effectiveness of the proposed framework. The submitted system achieves a tcpMER of 15.41% and ranks second among all participating teams.

cs.CL↗

arXiv2026

The tttAI System for the TSA-ASR Task of the SmartGlasses Challenge 2026

This paper presents the tttAI system submitted to the TSA-ASR task of the SmartGlasses Challenge 2026, evaluated on both two-person dialogues (Track 1) and multi-party meetings (Track 2). The task requires time-stamped speaker-attributed speech recognition from smart-glasses recordings. This is particularly challenging due to long-form audio, multiple speakers, and frequent overlapping speech. We proposed a cascaded architecture consisting of speaker diarization, overlap detection, target-speaker extraction, post-processing, and automatic speech recognition. The diarization module extracts features via WavLM-Large, performs frame-wise speaker classification with a Conformer encoder, and then generates global speaker segments through embedding clustering. For overlapped regions, we apply a WeSep-based target-speaker extraction model with ECAPA-TDNN speaker embeddings. When the extraction is unreliable, a dominant-speaker fallback strategy is used. The final system uses FireRedASR2-AED with the first microphone channel. The submitted system has a total parameter count of approximately 1.53B. On Track 1, our system achieves a tcpCER of 7.10%. On Track 2, it achieves a tcpCER of 34.04% and ranks second on the leaderboard.

eess.AS↗

arXiv2026

A Trajectory-Driven Spatio-Temporal Refinement Solution for CVPR 2026 8th UG2+ Challenge Track 3: DOST

In this work, we present our solution for the 8th UG2+ Challenge (CVPR 2026) Track 3: Dynamic Object Segmentation in Turbulence (DOST). Our method is built upon the strong baseline framework Segment Any Motion (SegAnyMo), which provides powerful mask generation and motion tracking capabilities. To further boost the segmentation performance under severe atmospheric distortions, we propose two key improvements. First, we employ a data-centric domain adaptation strategy. We significantly expand our training data by incorporating selected sequences from the DAVIS dataset alongside a subset of the DOST dataset, and apply simulated atmospheric fluctuation degradations to enhance the model's robustness against complex geometric distortions. Second, we introduce a spatio-temporal post-processing module. This refinement step effectively removes persistent boundary-connected false foregrounds and short-lived fragmented noise, while strictly preserving genuine small targets and maintaining original individual labels across frames. With these combined strategies, our proposed method ranks the 2st place in the challenge.

cs.CV↗

arXiv2026

X-Restormer++: 1st Place Solution for the UG2+ CVPR 2026 All-Weather Restoration Challenge

In this work, we present our winning solution for the 8th UG2+ Challenge (CVPR 2026) Track 1: Image Restoration under All-weather Conditions. Our method is built upon the X-Restormer baseline, which captures both channel-wise global dependencies and spatially-local structural information through its dual-attention design (Multi-DConv Head Transposed Attention and Overlapping Cross-Attention), augmented with the spatially-adaptive input scaling mechanism from Restormer-Plus. We adopt a two-stage training strategy with dual-model ensemble inference. In the first stage, Model B is trained from scratch on a large-scale diverse dataset randomly sampled from the FoundIR training set (approximately 800 GB out of 4.84 TB), covering five degradation types: blur, haze, rain, snow, and composite conditions such as co-occurring rain and haze. In the second stage, Model A is fine-tuned on the WeatherStream dataset (rain and snow splits) using Model B's final checkpoint as pretrained initialization, enabling efficient domain adaptation with a substantially smaller dataset. To better preserve structural details during training, we propose a novel Gradient-Guided Edge-Aware (GGEA) Loss, which applies Sobel operators to the ground-truth image to construct a spatially adaptive weight map that assigns higher supervision to edge and high-frequency regions. This is incorporated alongside L1 and Multi-Scale SSIM losses in a unified training objective. At inference time, predictions from the two models are fused via a weighted average, out = 0.4 x outA + 0.6 x outB, where the higher weight assigned to Model B reflects its stronger generalization ability from large-scale pretraining. With these strategies, our proposed method successfully ranks 1st in the challenge.

cs.CV↗

arXiv2025

LSVOS 2025 Challenge Report: Recent Advances in Complex Video Object Segmentation

This report presents an overview of the 7th Large-scale Video Object Segmentation (LSVOS) Challenge held in conjunction with ICCV 2025. Besides the two traditional tracks of LSVOS that jointly target robustness in realistic video scenarios: Classic VOS (VOS), and Referring VOS (RVOS), the 2025 edition features a newly introduced track, Complex VOS (MOSEv2). Building upon prior insights, MOSEv2 substantially increases difficulty, introducing more challenging but realistic scenarios including denser small objects, frequent disappear/reappear events, severe occlusions, adverse weather and lighting, etc., pushing long-term consistency and generalization beyond curated benchmarks. The challenge retains standard ${J}$, $F$, and ${J\&F}$ metrics for VOS and RVOS, while MOSEv2 adopts ${J\&\dot{F}}$ as the primary ranking metric to better evaluate objects across scales and disappearance cases. We summarize datasets and protocols, highlight top-performing solutions, and distill emerging trends, such as the growing role of LLM/MLLM components and memory-aware propagation, aiming to chart future directions for resilient, language-aware video segmentation in the wild.

cs.CV↗

arXiv2025

Enhancing Sa2VA for Referent Video Object Segmentation: 2nd Solution for 7th LSVOS RVOS Track

Referential Video Object Segmentation (RVOS) aims to segment all objects in a video that match a given natural language description, bridging the gap between vision and language understanding. Recent work, such as Sa2VA, combines Large Language Models (LLMs) with SAM~2, leveraging the strong video reasoning capability of LLMs to guide video segmentation. In this work, we present a training-free framework that substantially improves Sa2VA's performance on the RVOS task. Our method introduces two key components: (1) a Video-Language Checker that explicitly verifies whether the subject and action described in the query actually appear in the video, thereby reducing false positives; and (2) a Key-Frame Sampler that adaptively selects informative frames to better capture both early object appearances and long-range temporal context. Without any additional training, our approach achieves a J&F score of 64.14% on the MeViS test set, ranking 2nd place in the RVOS track of the 7th LSVOS Challenge at ICCV 2025.

cs.CV↗

arXiv2025

Pseudo-Label Enhanced Cascaded Framework: 2nd Technical Report for LSVOS 2025 VOS Track

Complex Video Object Segmentation (VOS) presents significant challenges in accurately segmenting objects across frames, especially in the presence of small and similar targets, frequent occlusions, rapid motion, and complex interactions. In this report, we present our solution for the LSVOS 2025 VOS Track based on the SAM2 framework. We adopt a pseudo-labeling strategy during training: a trained SAM2 checkpoint is deployed within the SAM2Long framework to generate pseudo labels for the MOSE test set, which are then combined with existing data for further training. For inference, the SAM2Long framework is employed to obtain our primary segmentation results, while an open-source SeC model runs in parallel to produce complementary predictions. A cascaded decision mechanism dynamically integrates outputs from both models, exploiting the temporal stability of SAM2Long and the concept-level robustness of SeC. Benefiting from pseudo-label training and cascaded multi-model inference, our approach achieves a J\&F score of 0.8616 on the MOSE test set -- +1.4 points over our SAM2Long baseline -- securing the 2nd place in the LSVOS 2025 VOS Track, and demonstrating strong robustness and accuracy in long, complex video segmentation scenarios.

cs.CV↗

arXiv2025

Transsion Multilingual Speech Recognition System for MLC-SLM 2025 Challenge

This paper presents the architecture and performance of a novel Multilingual Automatic Speech Recognition (ASR) system developed by the Transsion Speech Team for Track 1 of the MLC-SLM 2025 Challenge. The proposed system comprises three key components: 1) a frozen Whisper-large-v3 based speech encoder, leveraging large-scale pretraining to ensure robust acoustic feature extraction; 2) a trainable adaptor module using Linear-ReLU-Linear transformation mechanisms to effectively align speech and text representations; and 3) a frozen Qwen2.5-7B-Instruct large language model (LLM) integrated with trainable LoRA for optimized contextual linguistic decoding. By systematically combining pretrained models with task specific fine-tuning, the system achieved a word/character error rate (WER/CER) of 9.83% across 11 languages in the evaluation set and ranked third place among global participants.

eess.AS↗

arXiv2024

Frequency-selective terahertz wave amplification by a time-boundary-engineered Huygens metasurface

Ultrafast manipulation of optical resonance can establish the time-boundary effect in time-variant media leading to a new degree of freedom for coherent control of electromagnetic waves. Here, we demonstrate that a free-standing all dielectric Huygens metasurface of degenerate electric and magnetic resonances can prompt the broadband near-unity transmission in its static state, whereas it enables wave amplification in the presence of time boundary. The time boundary is realized by femtosecond laser excitations that transiently inject free carriers into the constituent meta-atoms for dynamic removal of a pre-established two-fold degeneracy. We observe that the transmittance in the photo-excited Huygens metasurface can exceed unity transmittance, i.e., THz wave amplification, by a factor over 20% in intensity at frequencies tunable by varying the arrival of time boundary with respect to that of the seed terahertz pulse. By numerical simulations and analysis with time-dependent coupled mode theory, we show that the wave amplification results from the ultrafast Q-switching and shift in resonant frequencies. This work demonstrates a new approach to achieve tunable amplification in an optical microcavity by exploiting the concept of time-variant media and the unique electromagnetic properties of Huygens metasurface.

physics.optics↗

arXiv2024

Multi-Modal Automatic Prosody Annotation with Contrastive Pretraining of SSWP

In expressive and controllable Text-to-Speech (TTS), explicit prosodic features significantly improve the naturalness and controllability of synthesised speech. However, manual prosody annotation is labor-intensive and inconsistent. To address this issue, a two-stage automatic annotation pipeline is novelly proposed in this paper. In the first stage, we use contrastive pretraining of Speech-Silence and Word-Punctuation (SSWP) pairs to enhance prosodic information in latent representations. In the second stage, we build a multi-modal prosody annotator, comprising pretrained encoders, a text-speech fusing scheme, and a sequence classifier. Experiments on English prosodic boundaries demonstrate that our method achieves state-of-the-art (SOTA) performance with 0.72 and 0.93 f1 score for Prosodic Word and Prosodic Phrase boundary respectively, while bearing remarkable robustness to data scarcity.

eess.AS↗

arXiv2023

Extremely thin perfect absorber by generalized multipole bianisotropic effect

Symmetry breaking plays a crucial role in understanding the fundamental physics underlying numerous physical phenomena, including the electromagnetic response in resonators, giving rise to intriguing effects such as directional light scattering, supercavity lasing, and topologically protected states. In this work, we demonstrate that adding a small fraction of lossy metal (as low as $1\times10^{-6}$ in volume), to a lossless dielectric resonator breaks inversion symmetry thereby lifting its degeneracy, leading to a strong bianisotropic response. In the case of the metasurface composed of such resonators, this effect leads to unidirectional perfect absorption while maintaining nearly perfect reflection from the opposite direction. We have developed more general Onsager-Casimir relations for the polarizabilities of particle arrays, taking into account the contributions of quadrupoles, which shows that bianisotropy is not solely due to dipoles, but also involves high-order multipoles. Our experimental validation demonstrates an extremely thin terahertz-perfect absorber with a wavelength-to-thickness ratio of up to 25,000, where the material thickness is only 2% of the theoretical minimum thickness dictated by the fundamental limit. Our findings have significant implications for a variety of applications, including energy harvesting, thermal management, single-photon detection, and low-power directional emission.

physics.optics↗

arXiv2023

TranssionADD: A multi-frame reinforcement based sequence tagging model for audio deepfake detection

Thanks to recent advancements in end-to-end speech modeling technology, it has become increasingly feasible to imitate and clone a user`s voice. This leads to a significant challenge in differentiating between authentic and fabricated audio segments. To address the issue of user voice abuse and misuse, the second Audio Deepfake Detection Challenge (ADD 2023) aims to detect and analyze deepfake speech utterances. Specifically, Track 2, named the Manipulation Region Location (RL), aims to pinpoint the location of manipulated regions in audio, which can be present in both real and generated audio segments. We propose our novel TranssionADD system as a solution to the challenging problem of model robustness and audio segment outliers in the trace competition. Our system provides three unique contributions: 1) we adapt sequence tagging task for audio deepfake detection; 2) we improve model generalization by various data augmentation techniques; 3) we incorporate multi-frame detection (MFD) module to overcome limited representation provided by a single frame and use isolated-frame penalty (IFP) loss to handle outliers in segments. Our best submission achieved 2nd place in Track 2, demonstrating the effectiveness and robustness of our proposed system.

cs.SD↗