ZIA™ Showcase Overview
ZIA™ Showcase is a platform for demonstrating and benchmarking DMP's latest edge AI recognition models. ZIA™ Showcase supporting the latest AI recognition models and multiple hardware enables customers to efficiently and in real-time evaluate and verify the optimal combination of AI recognition models and hardware using their own datasets in an online environment, and contributes to save time and costs of developments of their applications.
ZIA™ SAFE regroups a set of modules that are used for driver assistance/analysis, while ZIA™ MOVE is dedicated to autonomous driving for robotic vehicles. While both are aiming for different target applications, they share common core technologies that can also be individually tested via ZIA™ Showcase portal.
ZIA™ SAFE Overview
ZIA™ SAFE is a development platform. Through professional services, it enables one to develop an advanced safe driving assistance system in a short period of time.
ZIA™ SAFE is made accessible via this portal for real time evaluation.
High-level functions, such as DMS (Driver Monitoring System) and ADAS (Advanced Driving Assistance System) consist of core functions which can be evaluated, individually or collectively as technological blocks. For instance, in a high-level ADAS function, core functions such as lane detection, traffic signs detection, and distance estimation are included, each of which can be tested individually.
Other functions of DMP Core technologies that are not included yet in any high-level functions can be also evaluated as candidates, for your application.
In addition to providing tests for different functions, tests on different hardware platforms are available. Current available hardware for testing includes Xilinx SoC, Nvidia Xavier and NX, and Snapdragon 800 family.
As a fair performance benchmark, processing time and power consumption are provided. DMP provides offline results, but real time testing is available after your login.
ZIA™ SAFE and Related Core TechnologiesHigh-level functions include DMP ZIA™ SAFE for DMS and ADAS for your selection.
DMS using in-vehicle camera images includes the following core functions:
- Drowsy driving detection
- Face orientation (inattentive driving) detection
- Smartphone (mobile phone) detection
- Hand position detection by detecting the skeleton of the arm (distracted driving detection by determining whether drivers have an object in their hand or not)
ADAS includes the following core functions:
- Forward vehicle collision detection, rear vehicle collision detection, side vehicle collision detection
- Forward start delay detection
- Traffic light ignoring detection
- Traffic light location and color detection
- Stop sign ignoring detection
- Road signs (railroad crossings, stop lines, stop signs, etc.) detection
- Various road signs can be detected by learning their images
- Stop detection at stop line
- Number plate recognition and mosaic processing
- Face mosaic
- Road structure recognition by segmentation analysis
- Target road structures are own lanes, sidewalks, roads, fences, poles, dividers, etc.
- Judgment whether to contact road structures from the positional relationship with the own lane
ZIA™ MOVE Overview
ZIA™ MOVE is a technology platform enabler for customers to implement their autonomous vehicles solution.
ZIA™ MOVE contains all the required modules to start an autonomous PoC.
ZIA™ MOVE is highly configurable and extendable:
- Multiple sensors supported: additional sensor can be easily added upon request
- Modular architecture: existing modules can be adapted to or new modules can be created for specific requirements
- Multiple HW platforms for edge supported: additional HW can be added (GPU not mandatory)
- Accuracy/performances trade-off upon request
ZIA™ MOVE is map-free:
- Autoware relies on map / previous knowledge
- UGV using ZIA™ MOVE can run in completely unknown environment
ZIA™ MOVE on ZIA™ showcase aims to provide a mean for the user to evaluate DMP autonomous pipeline. Although all modules are important, ZIA™ SLAM is the key module for successful autonomous driving. ZIA™ Showcase offers the possibility to try customer data and obtain a result similar to the above video. The data used for this example is based on KITTI dataset.
Perception modules can be added on-demand. See ZIA™ showcase for ZIA™ SAFE for examples. People, cars, wires and Optical Character Recognition (OCR) modules are compatible between ZIA™ SAFE and ZIA™ MOVE.
Steps for uploading a video is as follows:
- Prepare the data using the following script (Available after login)
- Upload your data using the menu below
- Visualize the results
Currently, ZIA™ MOVE on ZIA™ showcase focuses on SLAM only. Other modules, such as perception, trajectory or control are not available online, but available as part of DMP software solution package. Feel free to contact DMP for further details.
How to use ZIA™ Showcase
In order to fully use this portal, please make sure to loginAccess without a login is available and you can view results on pre-selected data as well as performance and power consumption data charts. However in that case, you will not be able to upload your own data for testing.
ZIA™ Showcase: Make your selection!
Driver Monitoring System
Monitor if a driver shows dangerous behavior
Advanced Driver Assistance System
Autonomous pipeline including SLAM, mapping and perception modules
Tiny Yolov4 object detection
Number Plate Detection
Deep learning-based Japanese number plate detection
Detect current lane
Face Authentication System
Face authentication demo
ADAS: Forward collision warning
ADAS: Traffic light detection
Behind Camera Warning
Behind Camera Warning demo
Depth from Video
Depth from Video
Detect power lines, fences and other wires