Index Catalog // Scholar@UC

Q2000 Deep Learning Model Package

Type:: Dataset
Description/Abstract:: Q2000 Deep Learning Model Package This Technical Resource Bundle provides the Q200 Deep Learning model for open access download and use. The Q2000 DL model is built to detect Maya structures in Lidar data visualized at one meter per pixel. Currently, this repository contains the ESRI ArcGIS compatible DL model in the format .dlpk. We expect to convert the model in its current form into Pytorch (.pt) and TensorFlow (.h5) formats to incude them also here for user access. The Q2000.dlpk file was created in 2025 at University of Cincinnati by Benjamin Britton, using ArcGIS Pro v.3.3 with data from the NASA Ames G-LiHT. It is intended as an experiment to evaluate the practicability of creating a broadscale deep learning model that can be used effectively to identify Maya structures in Lidar data across the length and breadth of the Yucatan Peninsula. The Q2000 model is the subject of an article, a draft of which is also included in this dataset, called "Evaluating Broadscale Deep Learning for Maya Settlement Detection in G-LiHT Lidar" which examines the process and rationale of the model development in detail. The article has been accepted for publication and this site is included in that article with a link to this permanent (DOI) publication site. To use the model with ArcGIS Pro, use a Lidar dataset converted to a 1m/pixel DEM file and visualized as a 3-channel RGB Hillshade or other customized visualization as source input. -Using the ArcGIS Pro Spatial Analyst Extension, the geoprocessing tool called "Detect Object Using Deep Learning" may be invoked. -For Input Raster, add your Lidar visualization (a Hillshade visualization might be easiest for most users). -For Output Detected Objects, specify a new Layer name, and this will be the layer on which the detections will be recorded and displayed. -For Model Definition, use Q2000.dlpk. -Unless you want to run "arguments", you can leave the Arguments boxes as their Default. -I suggest checking the box (On) for Non Max suppression because it will reduce the amount of overlapping detections if target objects are located very close to each other, and I suggest a Non-Maximum Suppression (NMS) ratio of 0.5. This will tend to suppress detections that overlap by more than 50 percent. -I suggest you use Pixel Space unchecked (Off), since it is for an unrelated sort of object detection. -Before you click run, open the "Environments" tab (at the top of the window, next to the "Parameters" tab). Leave all the settings at their defaults there - except scroll down to the bottom of Parameters tab to the section called "Processor Type", pull down the Processor type pull-down, and choose GPU (it is set to CPU by default). -Then click Run and it will generate a new layer showing its detections as bounding boxes around target objects. You can see details for each detection by opening the Attribute Table on the new layer. You can see a screen capture of such a configuration in the image called Q200DemoScreenCap.jpg, included in this site's dataset, showing a detection on G-LiHT transect Yucatan_South_GLAS_395 near Pixoyal, with a detection of a Maya staircase highlighted on the display, and its corresponding information highlighted in the Attribute Table for it.
Creator/Author:: Britton, Benjamin
Submitter:: Benjamin Britton
Date Uploaded:: 08/23/2025
Date Modified:: 11/13/2025
Date Created:: March 15, 2025
License:: Public Domain Mark 1.0

BUILD TO ADAPT: HOW E.D.E.N. REIMAGINES CONTINUOUS IMPROVEMENT FOR NONSTOP OPERATIONS

Type:: Student Work
Description/Abstract:: This capstone report presents E.D.E.N. (Every Day, Every Night) — an original continuous improvement framework designed for nonstop, high-dependency operations in industries such as logistics, aviation, digital infrastructure, and healthcare. Drawing on principles from Lean, Six Sigma, high-reliability organizations (HROs), and data-driven decision science, the paper introduces four interlocking pillars: Engaged & Empowered Teams, Data-Driven Continuous Feedback, End-to-End Alignment, and Nonstop Adaptive Resilience. Through detailed analysis of recent global disruptions — including the 2025 CrowdStrike outage, Boeing’s manufacturing failures, the Red Sea shipping crisis, and Taiwan’s semiconductor challenges — the work demonstrates how organizations can embed real-time adaptability, resilience, and continuous improvement directly into their operations. The E.D.E.N. framework is proposed as a new model for achieving operational excellence and resilience in an era where downtime is no longer an option.
Creator/Author:: Agoba, EJ
Submitter:: EJ Agoba
Date Uploaded:: 07/16/2025
Date Modified:: 07/16/2025
Date Created:: 2025-04-10
License:: Attribution-NonCommercial-ShareAlike 4.0 International

A versatile near-infrared fluorescent probe for fast assessment of lysosomal status via a large multimodal model

Type:: Image
Description/Abstract:: These are original images for a new paper. This paper will be published soon.
Creator/Author:: Chen, Rui
Submitter:: Rui Chen
Date Uploaded:: 06/23/2025
Date Modified:: 06/23/2025
License:: Open Data Commons Open Database License (ODbL)