OPPO has invested a lot in its camera tech and that shows in the smartphones of the brand. Adding to that laurel, OPPO has bagged the first-place and two-third place awards at the Computer Vision and Pattern Recognition (CVPR) Conference 2020. The OEM showed-off its prowess at the event, revealing the new department of Intelligent Perception and Interaction for future AI technologies. It won the awards in perceptual extreme super-resolution, visual localization, and human activity recognition. Let’s have an insight into these three technologies.
Perceptual extreme super-resolution
OPPO Research team demonstrated how to get over the issue of perceptual extreme super-resolution for an image having 16 magnification levels. The challenge required 280 participants to come-up with AI-driven solution to produce high-resolution images having the best perceptual quality. OPPO used the RFB-ERSGAN neural network for overcoming this hurdle. In smartphones this will be able to produce high-res images with amazing details.
Visual localization for handheld devices
For this challenge, OPPO came up with a monocular visual localization pipeline for assessing semantic and depth of an image. This comes handy for camera-driven augmented reality applications and pinpoint navigation in phones. In visual localization technology, OPPO bagged the first place for outdoor visual localization and the third place for indoor visual localization.
Activity detection in extended videos
The third place in the Activity Detection in Extended Videos challenge also dropped in OPPO’s bag. The challenge demanded participants to come-up with automatic activity detection algorithms to identify characters in the video frame and recognize human movement in complex environments. For smartphones this means a better way to identify moving objects for accurate lens focus.
