Behavioural labelling of Dynamic Objects using DL based Semi-Automated tool
- On a busy roads there are various objects both static/stationery and dynamic
- Dynamic objects can be people walking, different vehicles & cycles moving on road, kids playing and others
- Like human brain, an autonomous vehicle should should identify the objects, categorise them and understand the behaviour of these objects
- Consider a situation as below (image without labelling) – where there are no. of vehicles, pedestrians and cyclists.
- <Before-Image> – should be without labelling
- KPIT’s semi automated tool chain is used for behavior labelling of these objects.
- The tool, identifies various objects, then categories them into different types of vehicles like Passenger car, truck; pedestrians, cycles, others.
- Then the tool also assign behavior labels to these objects depending on the movement & direction of movement of these objects.
- A sample image is as below
- The semi automated tool(build using Deep Learning), annotates and labels the objects automatically and requires minimum intervention from engineers
- Precision achieved – 1-3 pixels
- Objects labels – 4 objects
- No. of frames labelled – ??
We are the First Company in India to get SPICE Level 3 certification in 2004 & World’s second company to be accredited the Automotive SPICE™ Level 5 in 2007
Automotive SPICE is an initiative by Automotive Special Interest Group (SIG), SPICE User Group and Procurement Forum with 10 renowned OEMs as its members. It is used by OEMs to evaluate the quality & process competencies as well as organizational maturity of software service providers – who cater to automotive corporations
The genesis of Auto SPICE is attributed to the consensus among a number of auto OEMs around the world, on the growing criticality of software components in motor vehicles and the need for stringent quality standard to be adhered by the supplier. As most of the vehicle functionality is controlled by its software content, its malfunction has gained much significance/concern. Not only the manufacturers, but also the software service providers are taking pro-active actions to address such issues. Auto OEMs are adopting evaluation procedures for quality of delivered embedded software in automotive system by performing process capability assessments of their suppliers and making provision for contractual demands with respect to software quality. Therefore, the value that this certification has brought to all stakeholders is substantial.
Our expansive experiences coupled with strong process focus has helped the company meet stringent requirements of Automotive SPICE™. Not all automotive suppliers enjoy this accreditation.
At KPIT, it’s a way of life. It demonstrates our commitment to continuous process improvement and quality.
Adopting ASPICE has helped in –
- Better quality and reduced rework, which improves productivity, cost, and schedule
- Higher predictability of project costs and schedule adherence
- Improved visibility of issues – early and throughout the development lifecycle
Automotive SPICE L5 re-certification is done at regular intervals. KPIT continues to stay ahead of the game and reap organizational benefits