Adjusting Vision AI models and datasets for the Edge

Описание: The Edge AI Foundation recently organized a challenge, inviting data scientists, computer vision engineers, researchers and others to optimize and train vision models based on large open-source datasets. Different approaches were proposed to contestants, whether from the perspective of the models or from the angle of the large datasets.
Pandarasamy Arjunan (a.k.a. Samy) was in the models camp and joined me on the IoT Show to discuss his approach and learnings during the challenge.

🔗 Check out Samy's project at https://github.com/samy101/Wake_Visio...
🙏 This special episode is sponsored by the Edge AI Foundation https://www.edgeaifoundation.org/

📚 Chapters (Powered by https://timeskip.io/?ref=olivier)
00:00:00 - Intro: Edge AI
00:01:28 - IoT Show: Samy
00:02:49 - IoT Middleware System
00:04:42 - Edge AI and AIoT
00:06:18 - Image Classification ️
00:07:43 - Edge AI in the Picture ️
00:09:14 - Efficient AI Chips
00:10:38 - Model Pruning Use ️
00:12:15 - Methods Overview
00:13:48 - Challenge Approach
00:15:42 - Network Structure ️
00:17:15 - Architecture Work
00:18:54 - Transfer Learning
00:20:41 - Generalized Model
00:22:31 - Top Recommendations
00:24:05 - Year of Edge AI