samy101/bd3-building-defects-detection-dataset

How samy101/bd3-building-defects-detection-dataset works

BD3 is a building-defect image dataset and benchmarking research repository, not a deployed inspection application. Its documented workflow runs from collected and annotated images through preprocessing, augmentation, train/validation/test splitting, model benchmarking, and result analysis. The README documents the benchmark models and dataset sizes; individual notebook internals and their exact data handoffs were not sampled, so those links are shown as unverified where necessary.

Main entry points: output-analysis.ipynb Output analysis · train-val-split.py Dataset splitting · dataset_augument.ipynb Dataset augmentation · nn-alexnet.ipynb AlexNet training

Components 17

Dataset preparation

BD3 image dataset
Video frame extraction code/data-process/vid-img.py
Image inventory code/data-process/img_path.py
Image renaming code/data-process/rename.py
Image resizing code/data-process/resize512.py
Dataset augmentation code/data-augmentation/dataset_augument.ipynb
Dataset splitting code/train-test-split/train-val-split.py

Model benchmarking

AlexNet training code/model-train/alexnet/nn-alexnet.ipynb
MobileNetV2 training code/model-train/mobilenet/nn-MobilenetV2.ipynb
ResNet18 training code/model-train/resnet18/nn-ResNet-18.ipynb
VGG16 training code/model-train/vgg16/nn-VGG16.ipynb
ViT training code/model-train/vit/vt-model.ipynb

Results and evaluation

Output analysis code/output analysis/output-analysis.ipynb
Model comparisons Results/model_comparision/orig_results.csv
Class-wise reports Results/classification report/orig_best_report.tex
Confusion matrices Results/confusion matrix/orig_confu.tex

Other components

Dataset researcher

Connections 20

Dataset researcher → BD3 image dataset: collects and annotates README.md
Video frame extraction → BD3 image dataset: extracts frames for
Image inventory → BD3 image dataset: lists images for
Image renaming → BD3 image dataset: renames images for
Image resizing → BD3 image dataset: resizes images for
BD3 image dataset → Dataset augmentation: augments dataset README.md
Dataset augmentation → Dataset splitting: splits augmented data
Dataset splitting → AlexNet training: supplies benchmark split
Dataset splitting → MobileNetV2 training: supplies benchmark split
Dataset splitting → ResNet18 training: supplies benchmark split
Dataset splitting → VGG16 training: supplies benchmark split
Dataset splitting → ViT training: supplies benchmark split
AlexNet training → Output analysis: evaluates predictions
MobileNetV2 training → Output analysis: evaluates predictions
ResNet18 training → Output analysis: evaluates predictions
VGG16 training → Output analysis: evaluates predictions
ViT training → Output analysis: evaluates predictions
Output analysis → Model comparisons: compares models
Output analysis → Class-wise reports: reports class metrics
Output analysis → Confusion matrices: reports confusion

Drawn by GitDiagram on November 13, 2026 from the repository's file tree, README and source files. Read as Markdown ·