arXiv subjects

Meng Liang

Publications and source records attributed to Meng Liang.

4 recordsLinked to original sources

arXiv2026

FSVL: Taxonomic Fast-Slow Vision-to-Language Collaborative Framework for Interpretable Confusable Fish Species Identification and Knowledge Generation in the Middle Yangtze River

The Yangtze River Basin is a critical region for freshwater biodiversity conservation. The implementation of the Yangtze River Ten-Year Fishing Ban has increased the importance of effective aquatic biodiversity monitoring and species-level identification. However, fine-grained identification of closely related freshwater fish species remains challenging due to subtle morphological differences. Existing deep learning methods often struggle to balance classification accuracy and inference efficiency, while LLMs may generate unsupported information when producing specialized biological knowledge. To address these challenges, this study proposes FSVL, a fast-slow dual-pathway framework integrating visual recognition with LLM-based Retrieval-Augmented Generation (RAG). A lightweight ResNet50 fast path performs preliminary classification across 51 fish species in the Middle Yangtze River Basin and routes predefined groups of easily confused species to a slow path for refined identification. Guided by taxonomic criteria, the slow path uses YOLOv8n to detect key morphological regions, including the head, dorsal fin, and tail, leveraging morphometric and outline traits to improve discrimination among closely related species. The framework further integrates a retrieval-enhanced Qwen3-14B backend, using soft tokens and Low-Rank Adaptation (LoRA) for parameter-efficient domain knowledge generation. On the evaluated test set of 607 images, the fast classifier achieved 99.67% accuracy, whereas the cascade corrected the two errors, yielding 100.00% overall accuracy; further validation on independent specimens and external conditions remains needed. Across 255 knowledge generation samples, scientific name accuracy rises from 91.37% to 100.00%, genus accuracy from 90.20% to 98.82%, and ROUGE-L from 0.335 to 0.655, supporting intelligent biodiversity monitoring and fisheries management.

cs.CV↗

arXiv2026

THPL: A Vision-to-Language Decision Support Framework for Rainbow Trout Feeding Management in RAS

In Recirculating Aquaculture Systems (RAS), precision feeding is critical for minimizing costs and improving fish welfare. However, existing methods lack cognitive alignment between fish behaviors and management knowledge, impeding translation into executable, interpretable feeding decisions. To address this, we propose THPL, a generative feeding decision framework tailored for rainbow trout (Oncorhynchus mykiss) in RAS. First, Fishsort extracts trajectories to establish an Activity Coefficient (AC) quantifying feeding intensity. Second, a Hierarchical Behavior Encoder (HBE) models individual temporal progression and collective dynamics using Temporal and Set Transformers, transforming trajectory tensors into dual-evidence representations of explicit physical and implicit soft tokens. Finally, these tokens are integrated with environmental parameters, metadata, and expert rules to fine-tune an LLM via LoRA, followed by counterfactual multimodal Direct Preference Optimization (mDPO) to reinforce causal reasoning. Results show that AC exhibits a statistically significant monotonic positive correlation with expert-annotated feeding intensity (Spearman $ρ= 0.925$, $p < 0.001$). Ablations indicate that decision accuracy improves from 33.33% (text-only baseline) to 93.33% with dual-evidence tokens, confirming that continuous spatiotemporal tokens provide necessary physical grounding for LLMs. Compared with standard LoRA, counterfactual mDPO elevates decision accuracy from 93.33% to 96.67%, advances METEOR from 58.10% to 85.30%, reduces Self-BLEU-2 from 58.79% to 52.88%, and increases Distinct-3 from 6.68% to 7.81%, suppressing templating and actuation biases while reinforcing causal consistency and operational safety. Overall, by integrating continuous kinematics with LLM reasoning, this study provides a novel decision support paradigm for precision aquaculture.

cs.AI↗

arXiv2025

Fluorescence intensity correlations enable 3D imaging without sample rotations

Lensless X-ray imaging provides element-specific nanoscale insights into thick samples beyond the reach of conventional light and electron microscopy. Coherent diffraction imaging (CDI) methods, such as ptychographic tomography, can recover three-dimensional (3D) nanoscale structures but require extensive sample rotation, adding complexity to experiments. X-ray elastic-scattering patterns from a single sample orientation are highly directional and provide limited 3D information about the structure. In contrast to X-ray elastic scattering, X-ray fluorescence is emitted mostly isotropically. However, first-order spatial coherence has traditionally limited nanoscale fluorescence imaging to single-crystalline samples. Here, we demonstrate that intensity correlations of X-ray fluorescence excited by ultrashort X-ray pulses contain 3D structural information of non-periodic, stationary objects. In our experiment, we illuminated a vanadium foil within a sub-200 nm X-ray laser beam focus. Without changing the sample orientation, we recorded 16 distinct specimen projections using detector regions covering different photon incidence angles relative to the X-ray free-electron laser (FEL) beam. The projections varied systematically as the fluorescing volume was translated along an astigmatism, confirming that FEL-induced fluorescence reflects real-space structural changes. Our results establish a new approach for lensless 3D imaging of non-periodic specimens using fluorescence intensity correlations, with broad implications for materials science, chemistry, and nanotechnology.

physics.optics↗

arXiv2025

Girlhood Feminism as Soft Resistance: Affective Counterpublics and Algorithmic Negotiation on RedNote

This article explores how Chinese female users tactically mobilise platform features and hashtag practices to construct vernacular forms and an exclusive space of feminist resistance under algorithmic and cultural constraints. Focusing on the reappropriation of the hashtag Baby Supplementary Food (BSF), a female-dominated lifestyle app with over 300 million users, we analyse how users create a female-centered counterpublic through self-infantilisation, algorithmic play, and aesthetic withdrawal. Using the Computer-Assisted Learning and Measurement (CALM) framework, we analysed 1580 posts and propose the concept of girlhood feminism: an affective, culturally grounded form of soft resistance that refuses patriarchal life scripts without seeking direct confrontation or visibility. Rather than challenging censorship and misogyny directly, users rework platform affordances and domestic idioms to carve out emotional and symbolic spaces of dissent. Situated within the broader dynamics of East Asia's compressed modernity, this essay challenges liberal feminist paradigms grounded in confrontation and transparency. It advances a regionally grounded framework for understanding how gendered publics are navigated, negotiated, and quietly reimagined in algorithmically governed spaces.

cs.CY↗