mkocabas

####################### retinanet.py
def default_null_model(
    pyramid_feature_size=256,
    name='null_submodel'
):
    options = {
        'kernel_size' : 3,
        'strides'     : 1,
        'padding'     : 'same',
    }

mkocabas

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% In this tutorial, we are going to implement knn algorithm.              %
% Our aim is to see the most efficient implementation of knn.             %
% You have to implement knn in two differnt ways:                         %
%            1) with two loops                                            %
%            2) without any loop                                          %
%                                                                         %
% you have to report the computation times of both pathways.              %
% Note: the distance metric is �Euclidean�.                                %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

mkocabas

import numpy as np
from pycocotools.coco import COCO
import os
import math
import keras
from keras.models import Model
from keras.layers import Dense, Input, Flatten, Reshape
from keras.utils import plot_model
from random import shuffle

mkocabas

name: "1546"

layer {
  name: "data"
  type: "ImageSegData"
  top: "data"
  top: "label"
  #top: "data_dim"
  include {
    phase: TRAIN

mkocabas

from keras.optimizers import Optimizer
from keras import backend as K
from keras.legacy import interfaces


class MultiSGD(Optimizer):
    """
    Modified SGD with added support for learning multiplier for kernels and biases
    as suggested in: https://github.com/fchollet/keras/issues/5920

mkocabas

def focal_loss(gamma=2, alpha=0.75):
	def focal_loss_fixed(y_true, y_pred): # compatible with tf backend
        pt_1 = tf.where(tf.equal(y_true, 1), y_pred, tf.ones_like(y_pred))
        pt_0 = tf.where(tf.equal(y_true, 0), y_pred, tf.zeros_like(y_pred))
        return -K.sum(alpha * K.pow(1. - pt_1, gamma) * K.log(pt_1))-K.sum((1-alpha) * K.pow( pt_0, gamma) * K.log(1. - pt_0))
    return focal_loss_fixed

mkocabas

tb = TensorBoard(
        log_dir='./logs/{}'.format(args.exp_dir),
        histogram_freq=0,
        batch_size=args.batch_size,
        write_graph=True,
        write_batch_performance=True,
        write_grads=False,
        write_images=False,
        embeddings_freq=0,
        embeddings_layer_names=None,

mkocabas

# coding: utf-8

# In[ ]:

"""
Copyright (c) 2017, by the Authors: Amir H. Abdi
This software is freely available under the MIT Public License. 
Please see the License file in the root for details.

mkocabas

mkdir coco
cd coco
mkdir images
cd images

wget http://images.cocodataset.org/zips/train2017.zip
wget http://images.cocodataset.org/zips/val2017.zip
wget http://images.cocodataset.org/zips/test2017.zip
wget http://images.cocodataset.org/zips/unlabeled2017.zip

mkocabas

import keras
from keras.layers import Conv2D, UpSampling2D, BatchNormalization, Input, Add, MaxPool2D, Activation, Concatenate
from keras.models import Model
from utils.config import num_stages, num_joints

nFeatures = 256
nStack = 8
nModules = 1
nOutChannels = num_joints