Biometric Pattern Recognition TRIZ Cases

Machine learning models analyze image data to identify customer gestures and determine relevant object features based on proximity.

A video processing apparatus detects subjects and determines a main subject based on priority within partial areas.

An image processing device extracts frame images from moving video by analyzing subject presence and calculating evaluation values.

A machine learning module identifies and tracks specific participants within event video streams.

A main object determination apparatus selects candidates using feature points and verifies identity across multiple frames.

A low dimension pose space projects user LBP feature vectors into a trained manifold.

Deferred neural rendering generates actor replacements by merging source and target parameters, eliminating artifacts in regional video distribution.

A body part detector uses a classifier trained on single and multi-user images to identify anatomical features from depth data.

Trained AI algorithms isolate representative content from background data to reduce transmission volume.

Computer vision algorithms detect customer gestures to calculate referral scores, bypassing declining survey response rates.

Slicing insect regions reduces transmission volume while maintaining resolution for accurate neural network identification.

An automated system scores digital images and videos based on inclusion factors to generate curated highlight collections.

A hand biometric system generates digital fingerprints from captured images to enable secure authentication without physical contact.

Vision based LIDAR merges wide and narrow view images to resolve image resolution versus field of view trade-offs.

Ear scan biometric subsystem collects touch-point locations and device orientation angles to generate unique user profiles for mobile authentication.

Integrated RF power detector replaces bulky measurement equipment by mapping wireless charging fields for safe consumer operation.

Machine learning models analyze video streams to select accurate images, resolving the trade-off between update frequency and manual effort.

A 2D pose estimation method predicts pedestrian future positions using key point movement characteristics and Kalman filtering.

An autonomous delivery truck follows a mailman using sensors and control mechanisms to minimize walking distance while reducing labor intensity.

System segments recognition signals into permanent, transient, and ephemeral categories to maintain reliability when facial features are obscured.

A processing apparatus extracts multiple feature values from images to identify changes in appearance.

A processing apparatus extracts three-dimensional shape data from camera images to perform personal authentication without storing visual records.

Audio sampling across multiple electronic devices identifies the specific unit responsible for determining and executing user tasks.

An image recognizing apparatus classifies overlapping candidate areas to discriminate human body targets based on likelihoods and positional relations.

Neural networks extract features from reference images to generate parameter vectors for automated 3D character model construction.

A convolutional neural network fuses image and depth features to predict 3D reference points for object localization.