Cars and pedestrians detection and tracking from quadcopters
We used Faster R-CNN network architecture for detection. Modified SORT algorithm is being used for tracking. Multiple neural network architectures inspired by U-Net
2
We used a fully convolutional neural network architecture inspired by U-Net and created by NVIDIA
3
Recognition and reading of panel meters for a large steel plant in Russia
Date and time were recognized by template matching (Euclidean distance); for panel meters we used a custom convolutional neural network
People segmentation in videos
We used a fully convolutional neural network architecture inspired by U-Net
Video and webcam human pose detection
Tracking people on recordings from surveillance cameras in department stores
We used YOLO to find people in video frames and DeepSort to extract movement tracks and to compare people between frames
Write to us
Telegram
WhatsApp
Viber
VK
Messenger
Skype
Let's make something great together
hello@appliedml.io
Jl. Belimbing Sari No. GA, Banjar Tambyak, Pecatu, South Kuta, Badung Regency, Bali 80364