Sneak peek at the tech we've been developing with our partners https://t.co/jyyTtHqVfL to automate texture extraction from photos of consumer goods in a fully automated 4-step process: 3D reconstruction from a photo, segmentation, unwrapping and lighting corrections. @ZEG16271167
We are very excited to start a new project co-funded by the UK's innovation agency, @innovateuk. We will collaborate with https://t.co/qWA8ucAmUT to develop a novel framework that will automate the creation of stunning visuals for #ecommerce.
64+ threads machines are now mainstream and it would be nice if CPU meters in htop automatically grouped threads to avoid filling up the screen. With more threads, CPU meters will eventually fill up the entire screen. @hisham_hm https://t.co/j5uGGc3AdY
@Cyanapse was in the news today in @brightonargus, alongside @mnemoscene_ltd. We talked about some of the cutting-edge tech we're currently working on, including our photorealistic image filters that can change the lighting in pictures within 20 ms.
https://t.co/sPGJM49tUk
@oasace For anyone concerned with metadata tracking in images, the mat/mat2 anonymising utility (https://t.co/rYWRelA0gh) will strip the IPTC special instruction.
@hardmaru@cgnorthcutt Manufacturer recommendations for PSU wattage are too low for heavy use cases (e.g. deep learning training with heavy models) and will likely cause random crashes. See section ‘Practical example: System crashing when using the GPU’ at https://t.co/HVJsWANGxI
@hardmaru@cgnorthcutt Are you sure this power supply is sufficient for this 4-GPU setup? I'd expect a PSU over 1800-2000W to achieve full stability during training for this kind of setup. We previously had trouble scaling PSU wattage for just 2 GPUs in a high-end system and ended up with a 1300W PSU.
2/2 To confirm UK residence, additional documents are required, e.g. the 12 x 5 = 60 monthly bank statements covering the past five years, but there is a hard limit of 10 documents per application. That's not quite right. #Brexit@ukhomeoffice#EU@EU_Commission
1/2 #Brexit settled and pre-settled status verification appears to be broken for self-employed people. The system fails to verify UK residence based on National Insurance records.
So delighted for FuseBox residents @cyanapse to have been accepted on the @DigiCatapult Machine Intelligence Garage programme. Can't wait to see where the support and access to tech will take them.
We are very excited to be joining the Machine Intelligence Garage programme! Thanks @DigiCatapult for the support that will help us bring our innovation to the next stage!
Learn about Cyanapse, a participant of the latest Machine Intelligence Garage programme. @cyanapse develops software tools to improve the perception of images and #data, using methods based on #AI and neuroscience research: https://t.co/MNF25yFp9T
Nice visual experience with the @BrainAnalyser at @Digicatbrighton last night. It was like diving into brain circuits of vision. https://t.co/hptYP363Kk