Object Classification through Deconvolutional Neural Networks Student Carlos

Object Classification through Deconvolutional Neural Networks Student: Carlos Rubiano Mentor: Oliver Nina

Cha. Learn Looking at People Action/interaction spotting on RGB data Recognize actions using 235 performances of 11 action classes recorded and manually labeled in continuous RGB sequences of people performing natural isolated and collaborative behaviors

Video Classification with CNN Use CNN structure to classify actions in videos Using similar network that classifies CIFAR and takes inputs of 32 x 3 Also do this using Image. Net network and takes inputs of 256 x 3 Extract the frames from the videos Applied optical flow on the frames of the videos Resized the data for faster performance, and match inputs of the network Optical flow gives temporal information across two frames

Cha. Learn Looking at People Dataset Optical Flow

Model

Results from Cha. Learn Looking at People 2014 With automatic segmentation

Results on Validation Set Spatial DCNN Stacks Randomly Pretrained with Initialized Fine tuning 1 frame 0. 53 0. 57 3 frames 0. 55 0. 57 5 frames 0. 57 0. 6 Temporal DCNN Stacks Randomly Pretrained with Initialized Fine tuning 1 frame 0. 58 0. 5 3 frames 0. 54 - 5 frames 0. 54 - With manual segmentation

Also… Add: - motion boundary histograms - Dense SIFT - HOG