Evaluation of Resnet on Nvidia Jetson TX2

Keras implementation of Resnet using the on-board camera of TX2.The Signs dataset was used to train this model.However,this can be used as a generalized image clasiification model.

Demo Link:https://youtu.be/_xkCEabTec4

Getting Started

The neural network was trained on Nvidia Titan X GPU.This model was later used with nvidia Jetson TX2 Board. Opencv was used to capture images .

Prerequisites

1.Python 3.5
2.Tensorflow 1.5
3.Keras
4.Scikit Learn
5.Open CV 3.4.1

Theory

The details of this ResNet-50 model are:

Zero-padding pads the input with a pad of (3,3)
Stage 1:
The 2D Convolution has 64 filters of shape (7,7) and uses a stride of (2,2). Its name is "conv1".
BatchNorm is applied to the channels axis of the input.
MaxPooling uses a (3,3) window and a (2,2) stride.
Stage 2:
The convolutional block uses three set of filters of size [64,64,256], "f" is 3, "s" is 1 and the block is "a".
The 2 identity blocks use three set of filters of size [64,64,256], "f" is 3 and the blocks are "b" and "c".
Stage 3:
The convolutional block uses three set of filters of size [128,128,512], "f" is 3, "s" is 2 and the block is "a".
The 3 identity blocks use three set of filters of size [128,128,512], "f" is 3 and the blocks are "b", "c" and "d".
Stage 4:
The convolutional block uses three set of filters of size [256, 256, 1024], "f" is 3, "s" is 2 and the block is "a".
The 5 identity blocks use three set of filters of size [256, 256, 1024], "f" is 3 and the blocks are "b", "c", "d", "e" and "f".
Stage 5:
The convolutional block uses three set of filters of size [512, 512, 2048], "f" is 3, "s" is 2 and the block is "a".
The 2 identity blocks use three set of filters of size [512, 512, 2048], "f" is 3 and the blocks are "b" and "c".
The 2D Average Pooling uses a window of shape (2,2) and its name is "avg_pool".
The flatten doesn't have any hyperparameters or name.
The Fully Connected (Dense) layer reduces its input to the number of classes using a softmax activation. Its name should be 'fc' + str(classes).

Running the code

1.Directory Structure

---------------------------------
Resnet_on_Jetson
|-resnet_utils.py
|-resnet_50.py
|-resnet_small.py
|-test_on_cam.py
|-data-|
|       |-train
|       |-test
|-----------------------------------

2.Implementation on Camera Jetson

|-The Resnet 50 Model is really huge.
|-It made the jetson really slow as it cold not fit into he memory
|-We created another small model for the same in file resnet_small.py

The facenet aims to minimize this triplet loss.
Here f(A,P,N) stands for the embeddings of each of the input image

3.Results

Authors

• Siddharth Bhonge - Parser /Model - https://github.com/siddharthbhonge

Acknowledgments

• Andrew Ng | Deeplearning.ai
  
• the ResNet algorithm due to He et al. (2015). The implementation here also took significant inspiration and follows the structure given in the github repository of Francois Chollet:
  
• Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun - Deep Residual Learning for Image Recognition (2015)
  
• Francois Chollet's github repository: https://github.com/fchollet/deep-learning-models/blob/master/resnet50.py