From KNOW-HOW to WOW

Amazing deal or cheap counterfeit product? Sometimes it’s hard to tell. Orgify can authenticate pre-registered products in an instant. This Deep Dive episode offers a closer look at how registration and authentication work. We also look at how the technology enables new ways for brands to get in touch with customers who bought a pre-owned product.

Fake! Or is it? It's getting harder and harder to tell if something is real or not. This applies not only to images or facts, but also to physical goods such as consumer electronics, pharmaceuticals, or luxury watches. Bosch engineer Louis Tepper is part of a team that has developed a solution: Origify is able to scan and recognize the surface of an item based on a simple smartphone image. Melena and Shuko test it. In the process, they learn a lot about the luxury and vintage watch industry. They meet a young Danish watchmaker, Rune Bakkendorff, who builds watches from scratch. Origify, it turns out, could not only help consumers identify counterfeit products, but also help experts authenticate spare parts or industry identify parts too small for serial numbers or QR codes. What time is it? Time to listen to this episode!

www.linkedin.com/showcase/bosch-origify EN: www.bosch-origify.com DE: www.bosch-origify.com/de

"Close the door!" If you're saying that to your voice assistant instead of your housemates, you have a smart home. But which manufacturer's smart home devices are you using, and do they talk to each other? That question may be less relevant in the future, thanks to a relatively new standard called Matter. Bosch expert Michael Zuegel has been busy making our smart home products Matter-ready. He shares his insights on the technology with hosts Shuko and Melena. They also talk to Youtuber Chris West, who maintains a complex smart home setup with automations that make life easier - except when they go wrong. Now tell your voice assistant, "Play the episode!"

Generative AI or large vision-language models can improve advanced driver assistance systems (ADAS) in various ways. Dr. Liu Ren of Bosch Research in Sunnyvale, California, explains how foundation models can help label training data and analyze a bird’s eye view representation of a car’s various data sources to improve the functional performance of ADAS.

Papers: Xiaoqi Wang, Wenbin He, Xiwei Xuan, Clint Sebastian, Jorge Henrique Piazentin Ono, Xin Li, Sima Behpour, Thang Doan, Liang Gou, Han Wei Shen, Liu Ren. "USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation", at CVPR 2024.

Chenbin Pan, Burhan Yaman, Tommaso Nesti, Abhirup Mallik, Alessandro G. Allievi, Senem Velipasalar, Liu Ren. "VLP: Vision Language Planning for Autonomous Driving", at CVPR 2024.

Chenbin Pan, Burhan Yaman, Senem Velipasalar, Liu Ren. "CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow", at CVPR 2024.

Prompt battle! We've all tried giving generative AI chatbots instructions that yield useful results. Recently, GenAI has rapidly expanded across various industries. But can it help drive a car? Dr. Liu Ren, Vice President and Chief Scientist for scalable and assistive AI at Bosch Research, is confident it can. He and his team are at the forefront of integrating foundation models such as large language models or vision-language models with advanced driver assistance systems (ADAS). In this episode, Liu explains to our hosts, Shuko and Geoff, how the extensive knowledge contained in foundation models can enhance the perception, prediction, and planning capabilities of ADAS. What's even more impressive is that Bosch Research’s approach works despite the constrained resources typical in vehicle environments. As Liu shares his cutting-edge research, Shuko and Geoff go head-to-head in a challenge: Who can craft the cleverest prompts for GenAI?

Credits:

1. Chenbin Pan, Burhan Yaman, Tommaso Nesti, Abhirup Mallik, Alessandro G Allievi, Senem Velipasalar, Liu Ren. “VLP: Vision Language Planning for Autonomous Driving”, The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024 (CVPR 2024).
2. Xiaoqi Wang, Wenbin He, Xiwei Xuan, Clint Sebastian, Jorge Henrique Piazentin Ono, Xin Li, Sima Behpour, Thang Doan, Liang Gou, Han Wei Shen, Liu Ren. “USE: Universal Segment Embeddings for Open-Vocabulary Image Segmentation” , The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024 (CVPR 2024). 3.Chenbin Pan, Burhan Yaman, Senem Velipasalar, Liu Ren. “CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow”, The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024 (CVPR 2024).

More Bosch podcasts: From Know-how to Wow with Liu Ren, all about “Visual Analytics”: https://www.youtube.com/watch?v=CmR97fW_B3I Beyond Bosch: https://podtail.com/de/podcast/beyond-bosch/