Search
Results
-
Understanding and quantifying the factors determining potential worst-case heat anomalies
(2024)Over recent decades, many unprecedented heat extremes have been recorded, resulting in devastating impacts on the environment, society, and economy. However, while the proximal causes of moderate heatwaves are clear (anticyclonic large-scale flow situations, subsiding air and clear skies, dry soils, and amplified land-atmosphere interactions), the physical processes that differentiate the potential worst-case heatwaves from somewhat ...Master Thesis -
Quantify the Unquantifiable: Social Framework for a Regenerative Built Environment
(2024)Certification and rating systems aim to reduce harm. However, they still fall under a degenerating system, resulting in incremental changes. This highlights the need to shift towards regenerative systems to create a positive impact. Blockchain technology can address design structures and mental models highlighted by regenerative design principles. Therefore, this study explores how blockchain technology can facilitate the transition from ...Master Thesis -
-
Instance Segmentation, Body Part Parsing, and Pose Estimation of Human Figures in Pictorial Maps
(2023)Training and test data to idenfity silhouettes of human figures on pictorial maps (6964 different images) and to segment body parts and pose points of pictorial human figures (9503 different images)Dataset -
-
Detection of Pictorial Map Objects with Convolutional Neural Networks
(2023)Training and validation data to classify maps and non-maps (6200 images), to classify pictorial maps and non-pictorial maps (3000 images) as well as to detect sailing ships on historic maps (3201 bounding boxes on 525 maps)Dataset -
Detection of Pictorial Map Objects with Convolutional Neural Networks
(2023)keras/TensorFlow models and preprocessed data to classify maps and non-maps (Xception and InceptionResNetV2), to classify pictorial maps and non-pictorial maps (Xception and InceptionResNetV2) as well as to detect sailing ships on historic maps (Faster R-CNN and RetinaNet)Model -
Inferring Implicit 3D Representations from Human Figures on Pictorial Maps
(2023)Training, validation, test and result data to estimate 3D poses, to infer 3D body parts, to predict UV coordinates. to inpaint and enhance textures of pictorial human figures (3186 posed 3D models of humans, 3925 images of cartoonized humans, 3133 images of cartoonized human heads)Dataset