BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Physical Sciences Data Infrastructure - ECPv6.18.3//NONSGML v1.0//EN CALSCALE:GREGORIAN METHOD:PUBLISH X-ORIGINAL-URL:https://www.psdi.ac.uk X-WR-CALDESC:Events for Physical Sciences Data Infrastructure REFRESH-INTERVAL;VALUE=DURATION:PT1H X-Robots-Tag:noindex X-PUBLISHED-TTL:PT1H BEGIN:VTIMEZONE TZID:Europe/London BEGIN:DAYLIGHT TZOFFSETFROM:+0000 TZOFFSETTO:+0100 TZNAME:BST DTSTART:20250330T010000 END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0100 TZOFFSETTO:+0000 TZNAME:GMT DTSTART:20251026T010000 END:STANDARD BEGIN:DAYLIGHT TZOFFSETFROM:+0000 TZOFFSETTO:+0100 TZNAME:BST DTSTART:20260329T010000 END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0100 TZOFFSETTO:+0000 TZNAME:GMT DTSTART:20261025T010000 END:STANDARD BEGIN:DAYLIGHT TZOFFSETFROM:+0000 TZOFFSETTO:+0100 TZNAME:BST DTSTART:20270328T010000 END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0100 TZOFFSETTO:+0000 TZNAME:GMT DTSTART:20271031T010000 END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTART;TZID=Europe/London:20261210T140000 DTEND;TZID=Europe/London:20261210T150000 DTSTAMP:20261104T113034Z CREATED:20261022T110132Z LAST-MODIFIED:20261104T113034Z UID:38180-1796911200-1796914800@www.psdi.ac.uk SUMMARY:Webinar: From Project to Platform: New Resources on PSDI – Session 7 DESCRIPTION:Registration link: https://us06web.zoom.us/webinar/register/WN_rpE3uS6uT_Gtk8SMOxAA8Q \nPSDI is pleased to launch a new webinar series entitled “From Project to Platform: New Resources on PSDI”. This series aims to showcase the high-quality tools and resources developed through the funding call 2025\, introduce them to a broader community\, and foster engagement with relevant user groups.   \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n Abstract\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n TEMPOS – Simulate\, Train\, Deploy: A Reusable Workflow for Image Analysis with Limited Experimental Data at University College London. \n \n \n \n \n \n Challenge\n \n \n \n \n \n Machine learning has revolutionised image analysis; however\, it relies on extensive\, meticulously annotated training datasets that are frequently unavailable in experimental disciplines. In transmission electron microscopy (TEM)\, the quantification of nanoparticle size and morphology continues to depend predominantly on manual segmentation\, a process that is time-consuming\, subjective and difficult to scale. The annotation required to develop deep-learning models may exceed the duration of the experiments themselves. This labelling bottleneck impedes artificial intelligence adoption in any context where data acquisition is costly\, scarce or sensitive. \n \n \n \n \n \n Approach\n \n \n \n \n \n TEMPOS circumvents this constraint by utilising physically informed synthetic imagery rather than manually annotated experimental data. The methodology comprises three phases: \n\nSimulate: a generator produces realistic TEM micrographs with controllable particle size\, shape\, density and imaging conditions. Every image comes with exact ground truth\, known by construction.\nTrain: a Mask R-CNN instance-segmentation model\, built on Detectron2\, learns to outline individual particles from these synthetic images.\nDeploy: the trained model is applied to real micrographs. Per-particle morphological descriptors are extracted automatically and published to a FAIR\, searchable Datasette database.\n\nWe demonstrate the pipeline on latex spheres\, gold nanoparticles and Co₃O₄ nanocrystals\, and validate it against manually annotated experimental images. \n \n \n \n \n \n Innovation\n \n \n \n \n \n The central idea is that the training dataset is designed rather than collected. Labels come free with the simulation\, so new materials or imaging conditions can be handled by adjusting simulation parameters instead of running a new annotation campaign. The resulting datasets are AI-ready from the start. Combined with FAIR publication through the community database TEM-ParticlesDB\, TEMPOS offers a reusable template for any imaging problem where a reasonable physical model exists but labelled data do not. The webinar will walk through the workflow from start to finish and discuss how it can be adapted beyond electron microscopy. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n Dr Andrew Stewart directs the electron microscopy facilities within the Department of Chemistry at University College London\, where he oversees a multi-user electron microscopy facility (SEM & TEM) alongside an active research programme. His research focuses on electron diffraction techniques\, specifically three-dimensional electron diffraction (3DED/MicroED) and 4D-STEM\, and the integration of machine learning into electron microscopy and crystallography. A central objective of his work is enhancing the accessibility of these methodologies. This is achieved through open-source software and instrumentation\, notably the NanoMi\, TEMGYM\, and LiberTEM projects; the generation of synthetic training data for artificial intelligence\, and the development of FAIR community data infrastructure\, as exemplified by TEMPOS. He is a member of the IUCr Commission on Electron Crystallography and collaborates with PSDI and AIchemy.  \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n Register for this webinar\nRegister for this webinar directly through zoom:https://us06web.zoom.us/webinar/register/WN_rpE3uS6uT_Gtk8SMOxAA8Q  \n \n \n \n \n \n The PSDI team looks forward to seeing you at the webinar\, if you have any questions you can always get in contact with us. URL:https://www.psdi.ac.uk/event/new-resources-webinar-7/ LOCATION:Online\, Virtual Event\, Online CATEGORIES:Webinar END:VEVENT END:VCALENDAR