Welcome to JLUpub

Communities in JLUpub

Select a community to browse its collections.

Now showing 1 - 2 of 2

Recent Submissions

  • Thumbnail Image
    Item type:Item,
    Data, Dataset and Code Repository for AI-based shrimp fingerprinting study
    (2026-05-28) Bendag, Slim
    This repository contains python code that benchmarks local feature–based matching models against a classic computer vision baseline for shrimp re-identification across moulting events. The objective is to evaluate whether modern feature extractors and matchers can reliably re-identify the same shrimp before and after a moult. Each method is evaluated using two main criteria: • Re-identification accuracy • Percentage of shrimps correctly matched between moults. • Computational efficiency • Runtime per image pair and GPU usage. ***Repository structure** The repository is organized by method, with each folder containing a standalone fingerprinting pipeline: • d2-net/ – D2-Net feature extraction and shrimp fingerprinting pipeline • LoFTR/ – LoFTR-based matching pipeline • LightGlue/ – LightGlue + SuperPoint pipeline • SuperGluePretrainedNetwork/ – SuperGlue + SuperPoint pipeline • EDM/ – EDM model code and fingerprinting pipeline • xfeat/ – XFeat pipeline and training utilities • SIFT_OCV/ – SIFT baseline implemented using OpenCV Additional folders: • database/ – ROI cropped imaged dataset , organized by moult stage and shrimp ID • results/ – Output folders for each matching method • Stats/ – Jupyter notebooks, plots, and summary tables for analysis **Dataset layout** The dataset structure is: • database/moult1/<shrimp_id>/*.jpg • database/moult2/<shrimp_id>/*.jpg Notes: • Each <shrimp_id> directory may contain multiple images. • Each shrimp ID is treated as a class. • The scripts perform class-to-class matching, linking shrimp from moult 1 to moult 2. **Setup** 1. Create a Python 3.10 environment. 2. Install dependencies: 3. pip install -r requirements.txt 4. (Optional) Install a CUDA-enabled PyTorch build for GPU acceleration. The original setup used PyTorch 2.0.1 with CUDA 11.7. 5. Activate the virtual environment (venv). **Running the experiments** Each method has its own standalone script, with a configuration block at the top. Paths can be updated directly in the script. Example commands: SIFT • python SIFT_OCV/SIFT_OCV_fingerpritning.py LightGlue • python LightGlue/lightGlue_fingerprint.py LoFTR • python LoFTR/LoFTR_fingerprinting.py SuperGlue • python SuperGluePretrainedNetwork/superglue_fingerprinting.py D2-Net • python d2-net/D2-net_fingerprint.py EDM • python EDM/EDM_fingerprint.py XFeat • python xfeat/Xfeat.py **Outputs** Each method writes its results to a dedicated output folder, including: • score_matrix.csv – Class-to-class similarity matrix • mapping.csv – Hungarian assignment results with inlier counts and match rates • matched_keypoints.csv – Inlier correspondences for the best image pair per assigned class • vis/ – Visualizations of inlier matches after RANSAC filtering • run_resource_report.txt – Runtime statistics and GPU usage **Analysis and visualization** The Stats/ folder contains Jupyter notebooks that: • Aggregates results across all methods • Computes re-identification accuracy and efficiency metrics • Generates plots and CSV summary tables used in the final analysis
  • Thumbnail Image
    Item type:Item,
    Effect of recombinant fibroblast growth factor 10 on the lung vasculature in a mouse model of bronchopulmonary dysplasia
    (2025) Gersmann, Luisa
    There is strong evidence for the essential role of FGF10 during lung development. However, little is known about its concrete effect on the vascular system and the contribution of pulmonary vasculature to the pathogenesis of BPD. We aimed to study the impact of rFGF10 on the vasculature of the lung in a mouse model of BPD, to gain new information about underlying mechanisms and related pathologies such as BPD-PH. To demonstrate the effect of hyperoxia-induced lung injury, we exposed two experimental groups to hyperoxia (HYX) (PN1-PN8) followed by several intraperitoneal injections of PBS or rFGF10 until PN42. In the following experiments, we compared these groups to a control group held under normoxic conditions (NOX). Before lung harvest on PN45, we performed lung function and echocardiographic measurements to study organ function indices. We found strong indications for impaired right ventricular function represented by high TAPSE values in the HYX PBS group, suggesting an increased pressure of the pulmonary vascular system connected downstream of the right ventricle of the heart. The HYX FGF10 group showed a normalization of all studied indices in the functional experiments. Afterward, we collected lungs for further analysis involving RT-qPCR, immunostainings, and vascular morphometry after a double staining with ACTA2 and vWF. In summary, we found increased expression of Vegfa and its receptor Vegfr2 in the qPCR after rFGF10 treatment. The vascular morphometry revealed a rarefication of vessels in the distal areas of the lung after exposure to hyperoxia and regeneration after rFGF10 treatment. Furthermore, we were able to demonstrate an increased percentage of fully muscularized vessels towards a PH-phenotype after HYX lung injury and normalization after rFGF10 treatment in the periphery of the lung. Our results provide evidence for the affection of the pulmonary vasculature in a mouse model of BPD and the beneficial effects of rFGF10 treatment at a late stage of lung development. We found improved organ function, upregulated expression of pro-angiogenic genes and normalized histological features in the HYX FGF10 group compared to the HYX PBS group. In conclusion, this study reveals important information about changes of the pulmonary vasculature after hyperoxic lung injury at PN45, which represents young adulthood of the mice. We were able to partly reverse the damage described with our rFGF10 treatment. These findings might be crucial in developing targeted therapies to prevent or treat BPD and associated vascular comorbidities in prematurely born infants.
  • Thumbnail Image
    Item type:Item,
  • Thumbnail Image
    Item type:Item,
    Household expectations and dissent among policymakers
    (2025) Grebe, Moritz; Tillmann, Peter
    This paper studies the impact of dissent in the ECB’s Governing Council on uncertainty surrounding households’ inflation expectations. We conduct a randomized controlled trial using the Bundesbank Online Panel Households. Participants are provided with alternative information treatments concerning the vote in the Council, e.g. unanimity and dissent, and are asked to submit probabilistic inflation expectations. The results show that the vote is informative. Households revise their subjective inflation forecast after receiving information about the vote. Information about unanimity or dissent causes a wider individual distribution of future inflation for those households that were relatively certain before the treatment. For the remaining 60% of households, this information reduces uncertainty. Information about dissent increases uncertainty more than information about a unanimous vote, though the difference is not statistically significant. A unanimous vote unambiguously reduces inflation uncertainty for households with anchored inflation expectations.
  • Thumbnail Image
    Item type:Item,