Deeplite

Deeplite CEO, Nick Romano presents all things AI at the AI in Halton event!

Deeplite CEO Nick Romano talks #AI with the Greater Niagara Chamber of Commerce!

Deeplite's YOLOBench on Hugging Face Spaces is growing! Our latency-accuracy benchmark now includes over 900 YOLO-based object detectors! We've also added a feature to compare results across different hardware platforms to help to find the best YOLO model for your object detection use case! For you AI hardware providers out there if you'd like to have your hardware showcased on YOLOBench get your YOLOBench Hardware Kit here 👉 https://lnkd.in/gTY8RuYd

The proliferation of edge devices has unlocked unprecedented opportunities for deep learning model deployment in computer vision applications. However, these complex models require considerable power, memory and compute resources that are typically not available on edge platforms. Ultra low-bit quantization presents an attractive solution to this problem by scaling down the model weights and activations from 32-bit to less than 8-bit. DeepliteRT by Deeplite is an end-to-end solution for the compilation, tuning, and inference of ultra low-bit models on ARM devices. We implement highly optimized ultra low-bit convolution operators for ARM-based targets that outperform existing methods by up to 4.34x. 📣 You can read the full paper on arXiv ➡ https://lnkd.in/gjGWpSXk or for the TL;DR types you can check out our summary blog ➡ https://lnkd.in/grXfwBAZ Accepted at the British Machine Vision Association andSociety for Pattern Recognition BMVC2023 conference, You can watch out our poster presentation video as our very own Saad Ashfaq walks you through DeepliteRT!   ➡ https://lnkd.in/ghPQsz48 And lastly a shout out to the authors of this paper! Saad Ashfaq, Alexander Hoffman, Saptarshi Mitra, Sudhakar Sah (Sud), MohammadHossein Askari Hemmat, Ehsan Saboori (PhD, Eng) #machinelearning #ai #computervision #deeplearning #objectdetection

Deeplite is pleased to be partnering with Forum Ventures and Caden AI in hosting the next Generative AI meetup on November 16th! Caden AI Co-Founder and CEO Davis Sawyer and Deeplite Co-Founder and CTO Ehsan Saboori (PhD, Eng) will be diving into the technical challenges of adopting foundation and domain-specific gen AI models, such as creating data for fine-tuning, managing compute costs and hosting open source LLMs. The fireside chat will conclude with Q&A from the audience. This event is tailored to Founders, Funders, and Operators in AI and Tech. Join us and connect with some of the brightest minds in the AI industry, exchange fresh ideas, and foster invaluable support for your ventures. Event Details: 📅 Date: November 16th, 2023 ⏰ Time: 6:00 PM - 9:00 PM (Fireside starts at 7 PM) 🏠 Venue: (Provided upon registration confirmation) Spaces are Limited so sign up fast! Click here to register ➡ https://lnkd.in/egKEjTTH

Congratulations to Deeplite CEO Nick Romano on being named to the Canadian Chamber of Commerce (Canada) The Future of Artificial Intelligence Council!

Exciting news! We’re thrilled to announce the release of our interactive YOLOBench app on HuggingFace Spaces! 🚀 https://lnkd.in/gKimFAEm YOLOBench is a comprehensive latency-accuracy benchmark for over 550 YOLO-based object detectors, tailored for embedded use cases. 🚀 Use the YOLOBench app to find the best YOLO model for your edge device in no time! 🔍 Accepted at the ICCV 2023 RCV workshop, YOLOBench provides a fair comparison of models using a fixed training environment. It evaluates models on four datasets and five initial embedded hardware platforms, considering multiple dimensions of the model search space. 📊 Details can be found in our Medium article ➡ https://lnkd.in/gPDvHDWD Hey hardware providers! Want to showcase your AI hardware performance with YOLOBench users? Adding benchmark data for your devices is easy! Let’s make YOLOBench even better together! 💪 Get your YOLOBench Hardware Kit today! 📚🛠️ ➡ https://lnkd.in/gTY8RuYd #ai #yolo #objectdetection #deeplearning

Excellent job to the Deeplite team that worked so hard to get us here and Matteo Grimaldi for an awesome presentation at ICCV'23!

Congratulations to the Deeplite team and Matteo Grimaldi for an outstanding presentation at ICCV23 in Paris! We presented our posters for two papers accepted to the world renowned computer vision conference. YOLOBench: Benchmarking Efficient Object Detectors on Embedded Systems represents a benchmark comprised of 550+ YOLO-based object detection models on popular datasets & embedded hardware platforms. You can find the paper here ➡ https://lnkd.in/gBTD5iXs and the poster here ➡ https://lnkd.in/gtn93yVe _______ Accelerating Deep Neural Networks vis Semi-Structured Activation Sparcity presents a novel method to exploit sparsity in the network’s feature maps to reduce inference latency through minor runtime modifications. You can find the paper here ➡ https://lnkd.in/dbNgSH2e and the poster here ➡ https://lnkd.in/gZiTQa-Z