TDWG2026
Home / Talks / SAL A
talk · Friday 25 September · 10:30–⁠10:42 · SAL A

Automated Landmark Detection for Scalable Morphometric Data Extraction from Digitized Fish Collections

Bahadir Altintas · Large Language Models for Biodiversity Data Discovery, Integration, and Curation. AI applications, integration, and validation

Recording time 3:00:47–3:13:07Open on Vimeo ↗

Summaries and key moments are machine-produced. Check the recording for the speaker’s exact words.

The short versionA lightweight dlib shape-regression model can automate most fish landmarking from museum images, but inner landmarks and variable specimen poses remain hard.

Overview

What this was about

Bahadir Altintas presented a dlib-based machine-learning workflow for automatically placing 28 anatomical landmarks on images of Notropis minnows from museum collections. Fish are detected and cropped, landmarks are predicted, and predictions are compared with manual StereoMorph annotations. On 250 test images accuracy was 95% at a 0.02 error threshold and 85.3% at 0.01. Outline landmarks and the eye were most accurate; inner landmarks and curved, folded or dorsal-view specimens were hardest. Accuracy varied by species from about 66% to near 100%.

Why it matters. Manual landmarking limits morphometric studies. Automated landmarks could make digitised collection images an AI-ready source for large-scale shape analyses.

Key ideas

In the room

  • Dataset: 48 Notropis species, 28 landmarks from the literature, 217 training and 250 test images.
  • Manual annotation with the StereoMorph R package produced coordinates for training.
  • dlib was chosen for being fast, lightweight, CPU-efficient and extendable from face-landmark models; it uses regression to predict landmark positions.
  • Pipeline: detect the fish in the museum image, crop, predict landmarks.
  • Accuracy by Euclidean distance: 95% at 0.02 threshold, 85.3% at 0.01.
  • Outline landmarks were more accurate than inner-body ones; the eye was very accurate, perhaps because of its distinctive shape.
  • Per-species accuracy ranged from ~100% (N. chalybaeus) to 65.97% (N. buchanani), about 34% variation.
Jump in

Notable moments

In their words

Transcript

Automatically generated captions can contain mistakes, especially in names and technical terms. Times are relative to the room recording.

Read the transcript ↓
Loading transcript…
Keep exploring

Related talks.

See the map ↗
talk · 22 Sept · 12:03–⁠12:15 · SAL A

Towards Scalable Trait Measurements from Digitized Specimens

Matching rulers to an annotated reference set lets specimen images be calibrated to real-world units with under 1% error in 96% of cases.

Kenzo Milleville
talk · 24 Sept · 11:03–⁠11:15 · SAL B

Toward Quantitative AI-Readiness Metrics for Biodiversity Data Infrastructures

AI readiness should be a task-specific, measurable profile rather than a single score, and current biodiversity standards lack most of the terms needed to describe it.

Yasin Bakış
talk · 24 Sept · 11:51–⁠12:03 · SAL B

Provenance, Lineage, and Auditability in AI-Driven Biodiversity Image Workflows

Two persistent-identifier links per derived image, parent and batch, are enough to preserve auditable lineage and pipeline context for AI-processed biodiversity images, even outside the repository.

Xiaojun Wang
Programme abstract

Automated Landmark Detection for Scalable Morphometric Data Extraction from Digitized Fish Collections

Bahadir Altintas

Morphometric analysis is fundamental to studying biological shape variation, taxonomy, and biodiversity patterns, yet large-scale research remains constrained by the time and labor required for manual landmark annotation. As natural history collections increasingly digitize specimen images, the need for scalable and AI-ready approaches to morphological data extraction is rapidly increasing.

Here, we present a machine learning workflow for automated landmark detection in two-dimensional fish specimen images, built on a standardized anatomical annotation framework derived from established morphometric research. The model was trained and evaluated on manually annotated collection images to assess landmark detection accuracy across a diverse range of specimens.

Performance was evaluated using Euclidean distances between reference and predicted landmarks. Accuracy varied across anatomical regions, with external body outline landmarks achieving the highest precision. Species-level variation in performance was also observed, suggesting that morphology and imaging conditions both influence detection accuracy. Together, these results demonstrate that machine learning can reliably automate landmark detection in digitized collection images offering a scalable pathway toward AI-ready morphometric datasets for biodiversity research and natural history collections.

Submitted to the conference programme before the talk, so the talk itself may differ.