VIP20
本课程属于：005-Artificial Intelligence and Machine Learning Engineering English Technical Courses · 共 20 门相关课程，组合学习更完整。
查看所属VIP全部课程 →
← 返回所属目录 咨询 / 购买

Computer Vision Inspection Systems with Deep Learning

30 Chapters
View Course Samples Online
01
Inspection system architecture
imaging chain, lighting, optics, sensors, and deep learning pipeline integration.
Inspectionarchitecture
02
Defect taxonomy and data collection
labeling strategies, class imbalance, and dataset curation for industrial inspection.
taxonomycollection
03
Image preprocessing and augmentation
normalization, denoising, geometric transforms, and domain-specific augmentation for inspection.
preprocessingaugmentation
04
CNN fundamentals for inspection
convolution, pooling, receptive fields, and feature extraction on texture and surface defects.
CNNinspection
05
Transfer learning and fine-tuning
pretrained backbones, feature reuse, and adaptation to small industrial datasets.
Transferfine-tuning
06
Object detection for defect localization
two-stage and one-stage detectors, anchor design, and evaluation with IoU and mAP.
detectionlocalization
07
Semantic and instance segmentation for pixel-level defect analysis
U-Net, Mask R-CNN, and boundary refinement.
Semanticinstance
08
Anomaly detection with deep learning
autoencoders, GANs, and self-supervised methods for unknown defect discovery.
detectionautoencoders
09
Classification of surface defects
multi-class and multi-label models, confidence calibration, and rejection options.
Classificationmulti-class
10
Handling class imbalance and rare defects
resampling, loss functions, and synthetic defect generation.
Handlingimbalance
11
Real-time inference and edge deployment
model compression, quantization, pruning, and hardware acceleration.
Real-timeinference
12
Explainability and uncertainty in inspection decisions
saliency maps, uncertainty estimation, and human-in-the-loop review.
Explainabilityuncertainty
13
3D vision for inspection
point clouds, depth sensing, and volumetric defect detection.
inspectionvolumetric
14
Multimodal inspection
combining RGB, infrared, thermal, and acoustic signals with deep learning fusion.
Multimodalinspection
15
Inspection of printed circuit boards
solder joint defects, component placement, and automated optical inspection.
Inspectioncomponent
16
Semiconductor wafer inspection
pattern defect detection, yield analysis, and deep learning on wafer maps.
Semiconductorinspection
17
Metal surface inspection
crack, scratch, and corrosion detection under varying illumination.
inspectioncorrosion
18
Textile and fabric inspection
weave pattern analysis, stain detection, and high-speed line integration.
inspectionanalysis
19
Glass and transparent material inspection
reflection suppression, edge detection, and internal defect identification.
transparentmaterial
20
Food and agricultural inspection
foreign object detection, ripeness grading, and quality sorting.
agriculturalinspection
21
Pharmaceutical inspection
tablet defect detection, blister packaging, and label verification.
Pharmaceuticalinspection
22
Automotive and aerospace inspection
weld quality, coating defects, and assembly verification.
Automotiveaerospace
23
Model training pipelines
data loaders, distributed training, hyperparameter tuning, and experiment tracking.
trainingpipelines
24
Evaluation metrics and validation protocols
precision, recall, F1, AUC, and cross-validation for inspection.
Evaluationvalidation
25
Deployment and MLOps for inspection systems
model serving, monitoring, drift detection, and retraining.
DeploymentMLOps
26
Integration with robotic and automated systems
pick-and-place, conveyor tracking, and closed-loop control.
Integrationautomated
27
Case study
end-to-end deep learning inspection system for a manufacturing line.
end-to-endinspection
28
Case study
deep learning for anomaly detection in precision machining.
detectionprecision
29
Case study
large-scale visual inspection in electronics assembly.
large-scaleinspection
30
Future trends and challenges
self-supervised learning, few-shot inspection, and trustworthy AI in industry.
challengesself-supervised