Scene Recognition TRIZ Cases
- Camera Image Data Correctness Validation for Automated Driving
Validates vehicle camera image data by cross-referencing features across multiple sensors to ensure reliable input for automated driving functions.
- Photo Location Assignment Using Electronic Data Analysis
A photo location determination system analyzes user electronic information items to assign capture coordinates.
- Latent Semantic Sampling for Diverse Vehicle Trajectory Prediction
Generative adversarial network uses latent semantic sampling to produce diverse vehicle trajectory predictions.
- Driver Fault Influence Vector Characterization Using Motion Vectors
Motion vectors from video frames generate influence weights for driver scoring without raw image storage.
- Convolutional Neural Network Intermediate Classifier for Processing Load Reduction
Intermediate classifiers evaluate feature maps within convolutional neural networks to filter non-relevant data early in the processing pipeline.
- Sub-pixel Depth Estimation via Class Likelihood Parabola
Learning device calculates class likelihoods to estimate sub-pixel depth values using a parabola curve function.
- Frame Locked GPU Rasters for Real-Time Anti-Collision Processing
A frame locked multi-raster system processes object macrocells using synchronized GPU units to accelerate collision detection calculations.
- Camera Object Tracking Using Multi-Model Kalman Filters
Multi-model Kalman filter resolves zero-speed observation divergence in camera-based driver assistance systems.
- Onboard Digital Map Evaluation for Automated Driving Safety
A vehicle system analyzes sensor and map data to calculate a reliability probability for digital map items.