Everything around us today is getting Artificially Intelligent. Whether it is a social media website that can detect your friends from a picture or a self-driving car. Such great inventions are happening because of advancements in AI and its eco-system. Computer vision is one such important fields in Artificial Intelligence.

Computer vision is an interdisciplinary field that deals with how computers can be made for gaining high-level understanding from digital images or videos. -Wikipedia

Object detection has many practical uses, for example Face detection, People Counting, Vehicle detection, Aerial image analysis, security, etc.

Object detection is a computer technology related to computer vision and image processing that deals with detecting instances of semantic objects of a certain class (such as humans, buildings, or cars) in digital images and videos. -Wikipedia

From a developer standpoint building modern object detection application and models is not a straightforward task. However, companies like Google, Facebook, NVIDIA, Honeywell etc (more companies) are doing extensive research and open sourcing their findings in the form of data sets and algorithms. Of course, there is also a great pool of open-source community contributions like, OpenCV on of the popular computer vision library. Other most popular object detection algorithms/methods are R-CNN, Fast-RCNN, Faster-RCNN, RetinaNet, SSD, and YOLO. However, these algorithms may not achieve accurate results in all cases, for a variety of reasons:

1. Varying conditions
2. Variable number of objects
3. Sizing
4. Modeling

Alright, enough of theory lets build an image predictions application!

Setup:

Download and Install python: python

Download and Install python: pip

Tensorflow 1.4.0

pip3 install --upgrade tensorflow

Numpy 1.13.1

 pip3 install numpy

SciPy 0.19.1

 pip3 install scipy

OpenCV

pip3 install opencv-python

Pillow

 pip3 install pillow

Matplotlib

 pip3 install matplotlib

h5py

 pip3 install h5py

Keras 2.x

 pip3 install keras

ImageAI

pip3 install https://github.com/OlafenwaMoses/ImageAI/releases/download/2.0.2/imageai-2.0.2-py3-none-any.whl

RetinaNet model: Download RetinaNet and save it in your Image detection working directory

Now, are set to write our object detection models and draw insights from it.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sat Jul 28 15:39:18 2018

@author: aerup
"""
from imageai.Prediction import ImagePrediction
import os

current_dir = os.getcwd()
image_predictions = ImagePrediction()
image_predictions.setModelTypeAsResNet()
image_predictions.setModelPath(os.path.join(current_dir, "resnet50_weights_tf_dim_ordering_tf_kernels.h5"))
image_predictions.loadModel()        
images_array = list(filter(lambda x: x.endswith(".jpeg") or x.endswith(".png"),os.listdir(current_dir)))
results_array = image_predictions.predictMultipleImages(images_array, result_count_per_image=10)   
def consolidate_prediction_results(results):
    list_of_prediction = []
    for each_result in results:
        predictions, percentage_probabilities = each_result["predictions"], each_result["percentage_probabilities"]
        list_of_prediction.append(predictions)
        for index in range(len(predictions)):
            print(predictions[index] , " : " , percentage_probabilities[index])
        print("______________________________")
    print('set_of_prediction:',set([item for items in list_of_prediction for item in items]))    
consolidate_prediction_results(results_array)

Test Images:

Result:

cheetah  :  94.0053403378
leopard  :  1.18889305741
llama  :  0.692976033315
hyena  :  0.613261619583
jaguar  :  0.468546384946
trailer_truck  :  0.27624249924
Arabian_camel  :  0.211876211688
dalmatian  :  0.164259644225
yurt  :  0.148427509703
tow_truck  :  0.125058658887

Result:

motor_scooter  :  41.4442330599
sports_car  :  35.464990139
crash_helmet  :  9.53662842512
racer  :  6.1446454376
go-kart  :  1.31217772141
moped  :  1.12367495894
car_wheel  :  0.590410782024
convertible  :  0.586880231276
backpack  :  0.563116045669
car_mirror  :  0.377902877517

set_of_prediction

set_of_prediction: {'sports_car', 'motor_scooter', 'Arabian_camel', 'hyena', 'llama', 'racer', 'car_wheel', 'leopard', 'moped', 'convertible', 'jaguar', 'dalmatian', 'yurt', 'tow_truck', 'trailer_truck', 'backpack', 'cheetah', 'go-kart', 'car_mirror', 'crash_helmet'}

Above is the aggregate set of predictions by which you can develop interested analytical and decision models. Data of this kind has endless application.

We can then further analyze or process these images to pinpoint distinct attributes. For example:

Result of object Detection:

My vision:

One real-time application, I can associate with my field of work (ag-tech/ag-science) is a drone flying over a farm, capturing real-time crop pictures(corn for example), running through a model to detect anomalies(ex.diseases) and alert the farmer, keeping him informed and ahead of the game. In addition to that, we will be capturing the myriad of data which can then be utilized for R&D purposes.

Conclusion

The applications of computer vision, object detection, image predictions are numerous. I am anticipating more research and applications in these fields, shaping a variety of domains like security, agriculture, manufacturing, terrain analysis, commercial drones etc!

Cheers and Happy Building 🤘

Avinash Erupaka

I am a technology leader, with experience driving all aspects of technology transformation, from strategy to future state architecture. For last 10 years I worked for major corporations building B2B, B2C & internal platforms. I worked in the capacity of a senior dev, tech lead, Platform architect with hand-on experience driving technology strategy enabling business strategy. My expertise is in disciplines like distributed cloud Architectures, Data engineering and analytics, web and mobile application development, IoT, automation, security by design , agile - devsecops practices. I love managing teams and solving complex problems. I love tech, teaching, traveling, and fitness level boxing. I have proficiency using tools like React.js, Redux, Bootstrap, Material Design for the front end. Node.js, Scala, Clojure and Java for the back end. I leverage AWS, GCP, AZURE, OCI cloud platforms and I am a Multi Cloud Certified Architect. Opinions are my own and not the views of my employer.