Recently, the mobile QQ formally launched the "High-energy Dance Room" function. The new function is based on the 'Limb Motion Tracking' technology that is exclusively supported by Tencent's AI Lab Computer Vision Center. It also incorporates the Qualcomm snapdragon neural processing engine Referred to as SNPE) SDK, will provide more personalized content and user experience for young people to socialize. Mobile QQ is one of the most popular mobile Internet social APPs in China, which is popular among young people. In Q3 2017, the number of monthly active QQ accounts for 652.9 million in smart mobile devices. The realization of mobile QQ 'High Energy Dance Room' Is the first collaboration between Qualcomm and Tencent in the field of artificial intelligence and the first successful application of SNPE in China's Internet industry.The cooperation between both parties has realized the entry of artificial intelligence technology and has brought the leading experience of artificial intelligence support to a larger scale, Among the younger, younger user community. Through the SNPE SDK, this function can directly run the corresponding artificial intelligence neural network on the mobile phone without the need of processing in the cloud, in particular, the user can directly record the dancing short video according to the screen prompt and share the interactive dance through the QQ social relationship chain Video Qualcomm SNPE offers a high-performance dance studio with a high-performance and energy-efficient operating environment that significantly reduces the time it takes to recognize human pose and allows users to enjoy a smoother, more fun dance experience. Running artificial intelligence algorithms on the terminal side offers many advantages over artificial intelligence running in the cloud, such as instant response, increased reliability, increased privacy protection, and efficient use of network bandwidth, etc. To make it easier for developers and OEMs Using heterogeneous computing on the endpoint, Qualcomm introduced SNPE in 2016 to enable OEMs and application developers to run their own nerves on snapdragon terminals such as smartphones, security cameras, automobiles and drones Network models, and without having to be connected to the cloud at all, provide deep learning-driven experiences such as style conversion and filters (augmented reality applications), scene detection, facial recognition, natural language understanding, object tracking and avoidance, gestures and text Recognition, etc. SNPE is available on the Snapdragon 600 and 800 Series mobile platforms and supports generic deep learning frameworks such as Caffe, Caffe2 and Tensorflow with support for custom layers including run-time software, libraries, APIs, Offline model conversion tools, sample code, documentation, and debugging and benchmarking tools. For example, Facebook has announced plans to integrate SNPE into the camera capabilities of its Facebook application to promote enhanced Caffe2 support, in addition to the mobile QQ 'High Energy Dance Room' Realistic (AR) Features Compared to the usual CPU implementation, Facebook can exploit SNPE to deliver up to 5X performance improvements based on Adreno GPUs for a smoother, seamless, and lifelike AR capture of photos and live video Application characteristics. In addition, several mobile terminals equipped with Snapdragon mobile platforms have also implemented more advanced terminal-side AI experiences with SNPE, for example, OPPO R11s makes use of the innovative algorithm model of Soup Slim's miniaturization in perfect collaboration with SNPE, Play Xiaolong 660 mobile platform GPU, DSP computing power, a substantial increase in R11s artificial intelligence applications run-time processing speed, while reducing power consumption to optimize the camera experience, for example, in non-networked state, the user can photo real-time Increase portraits and bokeh effects, etc. Similarly, thanks to the SNPE supported by the Snapdragon 835 mobile platform, one plus 5T intelligently matches the 128 features of a face, completing a facial scan in just 0.4 seconds and proceeding Face recognition to help users easily and quickly unlock the phone. The Snapdragon 845, which Qualcomm just introduced in early December this year, is Qualcomm's third-generation AI mobile platform, delivering nearly three times the AI overall performance improvement over its predecessor, System-on-Chip (SoC). The SNPE SDK, in addition to the Google TensorFlow and Facebook Caffe / Caffe2 frameworks, adds support for Tensorflow Lite and the new ONNX to make it easy for developers to use the framework of their choice, including Caffe2, CNTK and MxNet.
Recently, the mobile QQ formally launched the "High-energy Dance Room" function. The new function is based on the 'Limb Motion Tracking' technology that is exclusively supported by Tencent's AI Lab Computer Vision Center. It also incorporates the Qualcomm snapdragon neural processing engine Referred to as SNPE) SDK, will provide more personalized content and user experience for young people to socialize. Mobile QQ is one of the most popular mobile Internet social APPs in China, which is popular among young people. In Q3 2017, the number of monthly active QQ accounts for 652.9 million in smart mobile devices. The realization of mobile QQ 'High Energy Dance Room' Is the first collaboration between Qualcomm and Tencent in the field of artificial intelligence and the first successful application of SNPE in China's Internet industry.The cooperation between both parties has realized the entry of artificial intelligence technology and has brought the leading experience of artificial intelligence support to a larger scale, Among the younger, younger user community. Through the SNPE SDK, this function can directly run the corresponding artificial intelligence neural network on the mobile phone without the need of processing in the cloud, in particular, the user can directly record the dancing short video according to the screen prompt and share the interactive dance through the QQ social relationship chain Video Qualcomm SNPE offers a high-performance dance studio with a high-performance and energy-efficient operating environment that significantly reduces the time it takes to recognize human pose and allows users to enjoy a smoother, more fun dance experience. Running artificial intelligence algorithms on the terminal side offers many advantages over artificial intelligence running in the cloud, such as instant response, increased reliability, increased privacy protection, and efficient use of network bandwidth, etc. To make it easier for developers and OEMs Using heterogeneous computing on the endpoint, Qualcomm introduced SNPE in 2016 to enable OEMs and application developers to run their own nerves on snapdragon terminals such as smartphones, security cameras, automobiles and drones Network models, and without having to be connected to the cloud at all, provide deep learning-driven experiences such as style conversion and filters (augmented reality applications), scene detection, facial recognition, natural language understanding, object tracking and avoidance, gestures and text Recognition, etc. SNPE is available on the Snapdragon 600 and 800 Series mobile platforms and supports generic deep learning frameworks such as Caffe, Caffe2 and Tensorflow with support for custom layers including run-time software, libraries, APIs, Offline model conversion tools, sample code, documentation, and debugging and benchmarking tools. For example, Facebook has announced plans to integrate SNPE into the camera capabilities of its Facebook application to promote enhanced Caffe2 support, in addition to the mobile QQ 'High Energy Dance Room' Realistic (AR) Features Compared to the usual CPU implementation, Facebook can exploit SNPE to deliver up to 5X performance improvements based on Adreno GPUs for a smoother, seamless, and lifelike AR capture of photos and live video Application characteristics. In addition, several mobile terminals equipped with Snapdragon mobile platforms have also implemented more advanced terminal-side AI experiences with SNPE, for example, OPPO R11s makes use of the innovative algorithm model of Soup Slim's miniaturization in perfect collaboration with SNPE, Play Xiaolong 660 mobile platform GPU, DSP computing power, a substantial increase in R11s artificial intelligence applications run-time processing speed, while reducing power consumption to optimize the camera experience, for example, in non-networked state, the user can photo real-time Increase portraits and bokeh effects, etc. Similarly, thanks to the SNPE supported by the Snapdragon 835 mobile platform, one plus 5T intelligently matches the 128 features of a face, completing a facial scan in just 0.4 seconds and proceeding Face recognition to help users easily and quickly unlock the phone. The Snapdragon 845, which Qualcomm just introduced in early December this year, is Qualcomm's third-generation AI mobile platform, delivering nearly three times the AI overall performance improvement over its predecessor, System-on-Chip (SoC). The SNPE SDK, in addition to the Google TensorFlow and Facebook Caffe / Caffe2 frameworks, adds support for Tensorflow Lite and the new ONNX to make it easy for developers to use the framework of their choice, including Caffe2, CNTK and MxNet.
|
