Motion classification using accelerometer, gyroscope and magnetometer sensors have been an important area of exploration for the past decade. Mostly studied in the context of health related applications, the implications of accurate inertial-magnetic motion classification span from continuous daily activity monitoring and remote assessment of patients recovery to athlete optimization and entertainment applications. While much has been done to optimize classification and segmentation algorithms, very little is understood of the effect sensor selection and motion representation has on overall system performance. In this paper, three sensors (accelerometer, gyroscope, orientation), seven motion representations and six classification techniques (K Nearest Neighbor, Artificial Neural Networks, Random Forests, Support Vector Machines, Naive Bayes) are compared. In addition to traditional time domain motion representations, a novel space domain representation is put forth which results in a two order of magnitude reduction in computational complexity. A case study dataset is created from 11 individuals performing 10 repetitions of 10 different upper body exercises. A single bicep mounted smart phone is used for data collection and both action classification and non-action rejection ability are studied.
This extended abstract describes our Ristretto framework for CNN compression. Last week Philipp presented the Ristretto Poster at ICLR’16 in San Juan.
High computational complexity hinders the widespread usage of Convolutional Neural Networks (CNNs), especially in mobile devices. Hardware accelerators are arguably the most promising approach for reducing both execution time and power consumption. One of the most important steps in accelerator development is hardware-oriented model approximation. In this paper we present Ristretto, a model approximation framework that analyzes a given CNN with respect to numerical resolution used in representing weights and outputs of convolutional and fully connected layers. Ristretto can condense models by using fixed point arithmetic and representation instead of floating point. Moreover, Ristretto fine-tunes the resulting fixed point network. Given a maximum error tolerance of 1%, Ristretto can successfully condense CaffeNet and SqueezeNet to 8-bit. The code for Ristretto is available.
Deep Convolutional Neural Networks (DCNN) have proven to be very effective in many pattern recognition applications, such as image classification and speech recognition. Due to their computational complexity, DCNNs demand implementations that utilize custom hardware accelerators to meet performance and energy-efficiency constraints. In this paper, we propose an FPGA-based accelerator architecture which leverages all sources of parallelism in DCNNs. We develop analytical feasibility and performance estimation models that take into account various design and platform parameters. We also present a design space exploration algorithm for obtaining the implementation with the highest performance on a given platform. Simulation results with a real-life DCNN demonstrate that our accelerator outperforms other competing approaches, which disregard some sources of parallelism in the application. Most notably, our accelerator runs 1.9X faster than the state-of-the-art DCNN accelerator on the same FPGA device.
Local Copy Download [PDF]
Presentation Download [PDF]
Link on IEEE Xplore
In this post I’ll talk in detail about the forward path implementation of the famous AlexNet. This Convolutional Neural Network (CNN) by Krizhevsky and Hinton has won the ILSVR 2012 competition with a remarkable margin. If you are starting to work with CNNs or Deep Learning in general, this post will give you a head start. You can find a straight forward implementation of the CNN’s forward path on our Github site. Feel free to download it and classify arbitrary images.
Continue reading “AlexNet Forward Path Implementation”
This document is the primary studies on the real.iZ VS-1000 3D Vision System for fruit detection. Please go to manuals and odos imaging for more information and always regard the manuals as your prioritized reference.
The contents are listed as follows:
- structure and operating principles
- indoor experiment
- outdoor experiments
- range photos processing
- portable solutions
- future challenges
- help seeking
Continue reading “Primary Studies on the real.iZ VS-1000 3D Vision System”
This document briefly explains how to use the parser of the Caffe code base in order to read and parse the specifications of neural networks. Caffe’s text file format for specifying models uses Google Protocol Buffer format. The code that actually reads a model can be found in src/caffe/util/io.cpp: Continue reading “Data Parser in Caffe”