Service scope and implementation plan
direction numberD58
Demand scenario8 items
main inputNIR L1;
Typical routeT11、T06、T07、T19、T20
The project scope is determined based on samples, equipment, working conditions and written indicators; general modules, special models and multi-sensor integration are configured according to the evaluation results.
Application scenarios and development content
IR457NIR Iris segmentation, quality evaluation and recognition
- Minimum input and technical route
- NIR L1；T11
- target output
- Iris area, quality and match score
- Implementation
- Need special model
- critical acceptance
- Stratified by distance/glasses/pupil, FMR/FNMR
IR458NIR Face detection and nighttime living assistance
- Minimum input and technical route
- NIR L3；T11
- target output
- Faces, key points and live features
- Implementation
- Security data required
- critical acceptance
- Attack presentation testing, skin tone/glass layering
IR459Thermal infrared face detection and cross-spectral identity research
- Minimum input and technical route
- LWIR+visible L4;T06+T07
- target output
- Face/identity candidates and confidence
- Implementation
- High risk assessment required
- critical acceptance
- Cross-device/environment ROC; must not be used by default for remote authentication
IR460Near-infrared palm vein/finger vein enhancement and identification
- Minimum input and technical route
- NIR L1；T11
- target output
- Vessel map, quality and match scores
- Implementation
- Requires special optics
- critical acceptance
- FMR/FNMR, living body/posture/skin color layering
IR461Eye movement, pupil and gaze NIR tracking
- Minimum input and technical route
- NIR L3；T11
- target output
- Eye Keys, Gaze and Blink
- Implementation
- Need special model
- critical acceptance
- Angle error, loss rate, glasses delamination
IR462Dark environment gesture and gesture interaction
- Minimum input and technical route
- NIR/LWIR L3；T07+T19
- target output
- Gestures/Postures and Events
- Implementation
- Product data required
- critical acceptance
- Gesture Macro-F1, time delay
IR463Low-resolution thermal array non-imaging gesture/presence sensing
- Minimum input and technical route
- Thermal array L3;T07
- target output
- Presence, direction and gross gestures
- Implementation
- Need low resolution model
- critical acceptance
- Event F1, multiplayer/background temperature layering
IR464Biometric data living body, attack prevention and privacy management
- Minimum input and technical route
- NIR/IR；T20
- target output
- In vivo results, risk and audit
- Implementation
- Security/compliance system required
- critical acceptance
- Attack coverage, deviation assessment, permissions and retention
Implementation and delivery methods
- Project input confirmationCheck camera, spectral band, bit depth, SDK, environmental conditions, sample and acceptance true value.
- Sample verificationVerify key processing links, applicable conditions and abnormal working conditions, and determine the development scope and acceptance methods.
- Algorithm development and integrationImplement data parsing, algorithms, quality gating, interfaces and target platform deployment.
- Test acceptance and handoverRetest according to the agreed test set and target platform, and hand over the deliverables, test results and applicable conditions.
- Requirements and technical solutions, input and output and interface field definitions
- Algorithm library, model or SDK/API; delivery form is confirmed in the project plan
- Sample program, parameter configuration, error status and calling instructions
- Test set version, indicator definition, stratification results and failed sample records
- Target platform adaptation, resource usage and running delay records
- Version, dependencies, license, reproduction steps and applicable conditions description
Acceptance design
Each scenario establishes independent test cases according to "key acceptance". When sharing models across scenarios, results are reported separately for each scenario, distance, load, environment, and target size, supplemented by overall statistics.
When high-risk alarms or controls are involved, the algorithm output is used for review or linkage judgment; safety functions, regulatory responsibilities, and automatic disposal require separate design and acceptance.
Frequently Asked Questions for Enterprise Customers
Can I purchase directly based on one of the scenes?
Can be evaluated first as a separate work package. The formal scope should describe the device, data, target objects, environment, interfaces, delivery form, and acceptance thresholds.
Can it be started when there are no abnormal samples?
The data interface, normal baseline, acquisition plan and simulation/controlled test design can be completed first; the abnormal detection rate and fault classification effect can be evaluated after obtaining sufficient abnormal true values.
Can the algorithm directly trigger alarms or equipment actions?
Duration, deduplication, quality gating, manual review, communication failure and safety interlocking need to be designed based on business risks. High-risk actions should be activated after completion of independent safety mechanisms and on-site acceptance.