Yolo + CoCo – openCv, Keras, Python, cuDnn
All posts by: michauPe
Testing serveral alghoritms to detect items in video using OpenCv, Keras, Python and CoCo / mobilenet . rcnn + keras mask yolo ssd mobilenet
Custom installation made for internal xmas party, done with use of budget rpis and standard cameras. Animated GIF – as a result – had got good quality versus price, very accurate synchro time. White ballance was a bit challenging and IMHO it requires better selection of lens. I have used 5 rpis 3B with standard cameras, one server on macbook with ubuntu – all connected in LAN via cables. The most important think is to have all cameras calibrated in center point. All software was wrote in Python and is easy extensiblle to large number of cameras. Basic x-flo lights were used to have 1/250 shutter speed. See You next Xmas !
First experiments with POV – powered by Arduino and 4 strips of LED WS. Using brushless motor, basic slip ring, and 3D printed stator. Funny how shutter speed of smartphone camera affects the optical illusion. Now it’s time write some piece of software to show some picture 🙂
1, 2, 3 !
4 units for rbk club c dj table almost done !
Workout challenge with Ewa Chodakowka. 3 rounds of exercises, made with magic mirror and unity3D. My role in project – concept and creative tech. direction.
Making off – shoot 360 in Tokio.
Few photos made during shooting 360 in Tokio for adidas.