According to reports from Electronic Enthusiasts, AI toys and AI smart glasses have become hot consumer electronics products this year, gaining momentum. We continue to track the dynamics of various chip manufacturers. Recently, Allwinner Technology has launched its V821 smart vision solution for AI smart glasses, which has completed its first customer release.This chip can also be used in the AI toy market and the security market, among others.The Allwinner V821 smart vision chipfeaturesa dual RISC-V architecture, with the main core supporting a maximum frequency of 1.2GHz and the small core supporting a maximum frequency of 600MHz. The dual-core heterogeneous design can meet various power consumption scenarios, including full-speed operation, low-speed AOV operation, Wi-Fi keep-alive sleep, and hibernation standby, enabling millisecond-level system cold start image and video capture. Its computing power supports algorithms for human detection, face detection, and face recognition for edge-side image processing.This chipsupports direct connection of dual cameras and allows for multi-camera/multi-resolution expansion, with a single camera supporting a maximum of 8M image sensors, meeting the edge-side upload requirements for various image clarity of current and future multimodal models. It supports professional-grade H.264 format at a maximum of 2K@30fps / 1080P@60fps, 8M MJPEG format hardware encoding, and supports MJPEG and H.264 decoding, meeting the low bandwidth and low latency requirements for cloud video-related services. It supports high-definition voice input, paired with various audio front-end algorithms to meet the low-latency real-time interaction needs between edge and cloud. It supports various display solutions, including a maximum of 800*480 RGB screen and dual SPI screens, catering to touch interaction, expression interaction, and audio-video material display functionalities.It can perform microsecond-level control processing based on the built-in RISC-V small core combined with RTOS, with multiple PWM interfaces to meet multi-motor control needs. It supports storage media such as SPI NOR/SPI NAND/SD NAND/eMMC, which can be used in combination (e.g., SPI NOR + SD NAND), meeting the expansion needs for various visual products and multimodal contextual memory products. It has built-in 2.4G Wi-Fi connectivity, allowing for outdoor direct video connection without dropping.The V821 has a package size of only 9mm*9mm, enabling extremely compact hardware design. With the development product package, the V821 can quickly meet the mass production needs of application solutions and can be paired with peripheral products to further achieve more innovative product expansions.For AI toys, Allwinner focuses on solvingkey issues such as chip computing power and power consumption optimization in AI toy scenarios.The Allwinner smart imaging chip integrates a self-developed AI ISP smart image processing engine, transplanting security-level image processing algorithms into consumer-grade AIoT products, achieving excellent image optimization effects under 4K high-definition images. Combined with visual modules and edge-side NPU computing power support, it can easily deploy AI algorithms such as OCR, picture book recognition, and sitting posture recognition, effectively improving the accuracy of visual functions such as object recognition, face recognition, and text recognition.Low power consumption isbased on years of algorithm iteration experience in the AIoT field, allowing Allwinner to accurately deploy low-power algorithms for different application scenarios. The multi-core heterogeneous design can achieve millisecond-level system cold start image and video capture functions while having a minimum power consumption of 3mA for VAD (voice activity detection), meeting various power consumption scenarios such as low-speed AOV (video always online) operation, Wi-Fi keep-alive sleep, and hibernation standby.Additionally,Allwinner provides a highly integrated algorithm toolkit, one-stop turnkey services, and an open platform. Leveraging the deployment advantages of AI large models like Alibaba Tongyi in the cloud, Allwinner helps manufacturers quickly realize the implementation of AI toy products.In the field of AI smart glasses,the Allwinner V821 flexibly adapts to various Bluetooth headset solutions, offering more choices in Bluetooth configuration. Based on Bluetooth wearable chips, it can seamlessly expand display functions; during high-definition image/video transmission, the V821’s high-speed bus can directly send high-definition video and images to mobile phones via built-in Wi-Fi, further saving power to extend battery life. The V821 has built-in low-power memory, allowing for more cost-effective storage options like SD NAND in addition to traditional solutions, further reducing overall costs.Based on the performance advantages and specifications of the V821 chip, Allwinner integrates years of experience in action camera image processing technology into a product package that meets the architecture of AI glasses, including chip layer, solution layer, and resource layer, facilitating the rapid expansion and implementation of visual wearable products, including AI glasses.For the small image sensors of 1/4~1/5 inch and small apertures used in glasses, targeted optimization is performed with professional-grade ISP. Adaptation has been completed for commonly used materials in AI glasses products, including: Geke Micro 5 million image sensors, Sony 8 million image sensors, and storage devices from Hengshuo and Xintianxia. More models are continuously being adapted based on demand.Furthermore,the next-generation V881 series chip, supporting 12 million to 20 million photo capture and 4K video recording, will debut in the second half of 2025.On the V821 wearable solution,furtherupgradesare underway.
Disclaimer: This article is reported by Electronic Enthusiasts, please indicate the source above when reprinting. For group communication, please add WeChat elecfans999, for submission and interview requests, please email [email protected].
More hot articles to read
-
Lenovo’s self-developed 5nm chip is here?! Suspected to be equipped with its own tablet
-
Price war, can gross margin still rise? These power semiconductor manufacturers made a fortune in the first quarter
-
HarmonyOS computer! Explosive debut!
-
Transformation! Domestic automotive chips are targeting the high-end market, with these products aiming for international standards
-
China’s AI market is expected to reach $50 billion! Jensen Huang: Missing the Chinese market will result in huge losses
Click to follow and star us
Star us to not miss any updates!