NEW CASE STUDY

Wildlife Tracking Uruguay

Building an end-to-end AI pipeline for automated wildlife recognition in Uruguayan camera trap videos.

95.7%

Test Accuracy

17×

Faster Pipeline

5 Parts

Deep-Dive Series

Read the Series → View on GitHub

Tech Stack: MegaDetector · ByteTrack · ResNet50 · PyTorch · Claude Code

🎯 11 Species Classified

Other Projects

Exploring computer vision, geospatial analysis, and tabular models

Featured Work

Amazon Berkeley Objects

Iterated VGG-based classifiers over 826k Amazon listings with aggressive label curation and data augmentation.

Transfer learning · CNN · Data augmentation

Stellar Classification — SDSS17

Balanced spectroscopic classes from SDSS using resampling, feature engineering, and calibrated multiclass classifiers.

Astronomy · Feature engineering · Multiclass classification

Wildlife Tracking Uruguay: Building an AI Pipeline for Conservation (Part 1)

A 3-stage modular pipeline (MegaDetector → ByteTrack → ResNet50) for automated wildlife recognition in Uruguayan camera trap videos, achieving 95.7% accuracy across 11 species.

Oct 10, 2025 Computer Vision, Conservation AI

Amazon Berkeley Objects

Jul 9, 2024 Computer Vision

Stellar Classification - SDSS17

Feb 23, 2024 Classification

Wine - Predicting Class

Nov 4, 2023 Classification

San Francisco Crime Classification

Context

Oct 15, 2023 Classification

Titanic - Machine Learning from Disaster

Sep 26, 2023 Classification

Wildlife Tracking Uruguay

Other Projects

Featured Work

Amazon Berkeley Objects

Stellar Classification — SDSS17

Wildlife Tracking Uruguay: Building an AI Pipeline for Conservation (Part 1)

Amazon Berkeley Objects

Stellar Classification - SDSS17

Wine - Predicting Class

San Francisco Crime Classification

Titanic - Machine Learning from Disaster

Etiquetas populares