walkoncross

name: "NormFace-Resnet"
input: "data"
input_dim: 100
input_dim: 3
input_dim: 112
input_dim: 96
layer {
  name: "flip_data"
  type: "Flip"
  bottom: "data"

walkoncross

name: "CenterFace-ResNet"
layer {
  name: "data"
  type: "ImageData"
  top: "data"
  top: "label"
  include { 
    phase: TRAIN
  }
  transform_param {

walkoncross

CenterFace (centerloss)

1. deploy:

• Netscope: http://ethereon.github.io/netscope/#/gist/719d1a0b2ae700598313d6833b04aaac

• Netscope CNN Analyzer: http://dgschwend.github.io/netscope/#/gist/719d1a0b2ae700598313d6833b04aaac

2. train-test

• Netscope:

walkoncross

Netscope: http://ethereon.github.io/netscope/quickstart.html

Netscope CNN Analyzer: https://dgschwend.github.io/netscope/quickstart.html

walkoncross

name: "CenterFace_Resnet"

input: "data"
input_dim: 1
input_dim: 3
input_dim: 112
input_dim: 96
layer {
  name: "conv1a"
  type: "Convolution"

walkoncross

(from https://www.scivision.co/numpy-image-bgr-to-rgb/)

Numpy / OpenCV image BGR to RGB

Conversion between any/all of BGR, RGB, and GBR may be necessary when working with Matplotlib expects M x N x 3 image, where last dimension is RGB.

OpenCV expects M x N x 3 image, where last dimension is BGR.

Scientific Cameras, some of which output an M X N x 3 image, where last dimension is GBR

walkoncross

import json
import os
import time
import requests
from PIL import Image
from StringIO import StringIO
from requests.exceptions import ConnectionError

def go(query, path):
  """Download full size images from Google image search.

walkoncross

How to write shared libraries

link: https://www.akkadia.org/drepper/dsohowto.pdf

walkoncross

	.file	"A.cpp"
	.section .rdata,"dr"
	.align 8
.LC0:
	.ascii "bits1: %5lu, bits2: %5lu, builtin: %5lu\12\0"
	.text
	.p2align 4,,15
	.def	_Z6printfPKcz.constprop.0;	.scl	3;	.type	32;	.endef
	.seh_proc	_Z6printfPKcz.constprop.0
_Z6printfPKcz.constprop.0:

walkoncross

__author__ = 'k0emt'

class Greeter:
    def __init__(self):
        self.message = 'Hello world'
        # print self.message