"Porting deep Neural Network for Cadence® Tensilica® AI Processor"

There are several deep neural networks of different types like CNN (Convolutional Neural Networks) or RNN (Recurrent Neural Networks) which are used in autonomous driving applications.

Cadence Tensilica Vision P6 DSP is a highly efficient neural and vision DSP, which can run implementations of CNN/RNN in an optimal way.

Compiling of a network to Vision P6 core is achieved using XNNC, a compiler tool which takes network description as input and generates implementation for Vision P6 core.

In this Cadence Tensilica Hackathon, we expect participants to take one of the networks from the list below and port using XNNC to achieve clean compilation and high efficiency of results.

• Classification Networks (e.g., MobileNet)
• Segmentation Networks (e.g., E-Net, SegNet, Shuffle Net)
• Object Detection Networks (e.g., Yolo, SSD, Faster and Masked RCNN)

Please note that the above list of networks is indicative and the organizers will provide a specific list two weeks before the Hackathon date.

Winning hacks will be judged based on objective performance numbers which are output by XNNC and also subjective evaluation of approach used by participants

Number of entries are limited to only 15 teams. You can form your own team of up to 3 members.

Awards will be presented at the Awards Ceremony on the evening of January 8, 2019.

Registration
Please send your entries by email with your names, name of institution to [email protected] by December 20, 2018.

Along with your entry email, include a short description about yourself and your team.

Schedule
Hackathon prep-up training – January 5 at 9.00AM sharp
Contest time –January 5, 12noon to January 6, 12noon

Venue
IIT Delhi Campus
Hauz Khas, New Delhi

What will be provided?

• Training on Cadence Tensilica XNNC (Xtensa Neural Network Compiler)
• Experts from Cadence will be around the entire time to help you during Hackthon
• All meals and snacks for the duration of the Hackathon
• One laptop per team

Winners will get exciting cash prizes:
1st place team cash prize of INR 65,000
2nd place team cash prize of INR 35,000