Protected book companion files, blog code utilities, and curated engineering resources for computer vision workflows.
Book companion resources are organized by track. Resources are provided via post-purchase email.
Companion files for the semantic segmentation book series, loss functions, metric helpers, mask utilities, U-Net references, architecture modules, and experiment support scripts.
Book ResourcesCompanion files for the object detection book series, dataset YAML templates, annotation checks, prediction helpers, evaluation utilities, ONNX export examples, and configuration references.
Book ResourcesCompanion resources accompanying published Occulins articles, including selected implementations, utilities, configurations, and supporting materials for practical experimentation and workflow understanding.
Supporting resources for segmentation debugging and foreground-collapse analysis, including U-Net implementation, loss functions, metric utilities, visualization helpers, foreground diagnostics, and experiment configuration.
Companion AssetsSupporting resources for custom YOLO training workflows on VisDrone, including dataset validation, YAML configuration, training utilities, prediction workflows, and evaluation guidance.
Companion AssetsSupporting resources for the U-Net architecture explanation, including model implementation, DoubleConv block, architecture configuration, skip-connection reference, and selected plotting utilities.
Companion AssetsSupporting resources for object detection failure analysis, including missing-label checks, YAML split verification, mAP comparison logic, overfitting diagnostics, and evaluation sanity checks.
Companion AssetsSupporting resources for handling class imbalance in segmentation, including dataset loading, loss function implementations (Dice, weighted BCE, focal loss), the shared U-Net architecture, training loop, and figure generation.
Companion AssetsSupporting resources for improving small object detection, including resolution training configurations, model scale comparisons, NMS threshold tuning notes, and a deployment trade-off guide.
Companion AssetsSupporting resources for evaluating object detection models, starting with the IoU computation used throughout the article. Precision/recall, PR curve, and confidence threshold utilities to follow.
Companion AssetsSupporting resources for understanding segmentation metrics, including a from-scratch per-class IoU, Dice, mIoU, sensitivity, and specificity implementation.
Companion AssetsSupporting resources for model deployment — TensorRT export, benchmark timing, and configuration reference from real YOLOv12 latency testing on RTX 4090.
Companion AssetsA practical checklist for catching validation overfitting, dataset bias, metric misinterpretation, and the qualitative-quantitative gap before deployment.
Companion AssetsPractical checklists for building a complete CV project — data auditing before training, model readiness before evaluation, and the key question to answer at each pipeline stage.
Companion AssetsAn ablation table template built from real VisDrone results, plus the limitation identification and ablation design frameworks used throughout this workflow.
Companion AssetsA realistic project timeline reference and the five questions every CV project should answer before work begins.
Companion AssetsCurated engineering resources for computer vision research, deep learning experimentation, edge AI deployment, robotics, UAV systems, and applied AI workflows.
| Resource | Best For | Technical Notes | Status |
|---|---|---|---|
| RTX 4090/5090 24GB | Heavy segmentation and detection training | Higher VRAM capacity supports larger input resolutions, heavier architectures, and larger batch sizes for deep learning experiments. | Coming Soon |
| RTX 4080/5080 16GB | Mid-to-high range experimentation | Useful for moderate segmentation and detection workloads with lower cost and power requirements than flagship GPUs. | Coming Soon |
| RTX 4070/5070 16GB | Lightweight to moderate AI research | Suitable for lightweight computer vision experimentation, architecture debugging, and deployment-focused model development. | Coming Soon |
| NVMe SSD 2TB | Dataset storage and fast experiment loading | Useful when working with image datasets, checkpoints, logs, and repeated training runs. | Coming Soon |
| Resource | Best For | Technical Notes | Status |
|---|---|---|---|
| Jetson Orin Nano | TensorRT FP16 deployment experiments | Useful for lightweight object detection and segmentation inference testing on embedded AI hardware. | Coming Soon |
| Jetson Orin NX | Higher-performance edge AI deployment | Suitable for heavier real-time computer vision workloads where embedded inference performance matters. | Coming Soon |
| Raspberry Pi AI Kit | Low-power AI prototyping | Useful for small embedded AI demonstrations and lightweight inference experiments. | Coming Soon |
| Resource | Best For | Technical Notes | Status |
|---|---|---|---|
| USB / RGB Camera Module | Real-time detection and segmentation demos | Useful for prototyping image acquisition pipelines and live computer vision experiments. | Coming Soon |
| Thermal Camera Module | Multimodal and environmental monitoring | Relevant for RGB-thermal object detection, fire monitoring, surveillance, and low-light perception workflows. | Coming Soon |
| AI Robotic Car Kit | Autonomous driving demonstrations | Useful for lane detection, obstacle detection, small-scale perception, and robotics vision experiments. | Coming Soon |
| UAV / Drone Platform | Aerial analytics and remote sensing experiments | Useful for UAV-based crack detection, aerial object detection, forest monitoring, and environmental inspection workflows. | Coming Soon |
| Resource | Best For | Technical Notes | Status |
|---|---|---|---|
| Deep Learning Reference Book | Foundational understanding | Useful for building stronger intuition around neural networks, optimization, and representation learning. | Coming Soon |
| Computer Vision Reference Book | Computer vision fundamentals | Useful for understanding classical and modern vision concepts before moving into applied deep learning systems. | Coming Soon |
| PyTorch Reference Book | Implementation and experimentation | Useful for researchers and engineers implementing models, training loops, debugging, and deployment pipelines. | Coming Soon |
Protected book resources may include selected model files, configuration files, modules, architecture references, and companion assets. Full proprietary repositories, commercial training frameworks, and complete production pipelines are not publicly distributed unless explicitly stated otherwise. Blog companion code may be released separately as open source where applicable.
Recommendations and external resource references are included for educational, research, and workflow guidance purposes. Some links may be affiliate links.