#!/usr/bin/env python
import tensorflow as tf
import torch
import rospy
import pandas as pd
from torch import nn
import matplotlib.pyplot as plt
mnist = tf.keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0
model = tf.keras.models.Sequential([
tf.keras.layers.Flatten(input_shape=(28, 28)),
tf.keras.layers.Dense(128, activation='relu'),
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(10)
])
We develop
AI and Robotics software
Willogy is a team of AI & Robotics researchers and software engineers
We work with start-ups, entrepreneurs, and business to level up their software products using Artificial Intelligence.
We help business owners to build innovative, market-fit products, MVP and optimize current solutions
Willogy Advantages?
What makes Willogy specical while there is a myriad of AI and software companies out there?
Brilliant-minded and Advanced-skill engineers
A+ engineers, researchers, and experts collaborating with Ph.D, PostDoc from Carnegie Mellon University (CMU), University of Adelaide, Monash University, National University of Singapore (NUS), Pohang University of Science and Technology (POSTECH), Singapore University of Technology and Design (SUTD), Imperial College London (ICL), Ho Chi Minh City University of Science (HCMUS)
Backed by scientific-proven and strong academia
Being advised and collaborating with professors from the University of Adelaide, SUTD, Monash University, Imperial College London Published scientific research in top conferences and journals in AI and Robotics for most of our products & solutions
Product-minded people for end-to-end product design and development
Blended with product-minded people who are skilled in successful (but also failed-to-almost-bankrupted) start-ups to understand the most abstract form of problems and ideas to shape up the good product-fit Leveraging Agile and Extreme Programming (XP) as guidance and practice to deliver less-risky products
Multi-platform development & deployment capabilities
Having solid background in both hardware and software to understand from the physical layer (the boring one-zero 10...0110 part) to the application layer, we are able to build not only cloud-based, web-based, mobile-based applications but also the Internet of Things (IoT), Robotics products, and Automation systems
Featured Case Studies
We have solved many challenging problems. These are some featured products we have built
PROBLEM
Optimize the surveillance of hand-washing process in surgery to minimize the risks of infection
SOLUTION
Build up a device with camera integrated to capture hand-washing process.
A dashboard running on web platform to view video analytics
Hand detection, Hand gesture recognition, Motion recognition, Motion detection, Video Analytics
TECHNOLOGY
Internet of Things Tensorflow Python Django C/C++
Hand Hygiene Compliance Surveillance in Surgery
Compliance with hand hygiene according to the World Health Organization (WHO) and C4.3 criteria is one of the quality assessment criteria in hospitals.
Every year, the Hospital Infection Control department monitors the rate and quality of hand hygiene compliance.
According to the assessment results, the average hand hygiene compliance rate in 2020 is > 70%.
However, the quality of the hand hygiene compliance cannot be measured.
Assessing the compliance of hand hygiene steps of medical staffs routine in clinical departments, thereby contributing to the development of a program to improve the quality of supervision and staff self-discipline
We have helped to develop an end-to-end system from hardward (IoT) devices to AI algorithm and analytics software on Web platform to empower the surveillance process of the hand hygiene compliance.
The system has saved a lot of human effort of the Infection Control department staffs to maintain the hand hygiene compliance rate and quality.
PROBLEM
The advanced wound care process need to be able to detect & measure anatomical and cellular structures for better treatment
SOLUTION
Reconstruct 3D model of a wound from a 2D video capture by mobile phone to identify wound size and depth
3D Reconstruction, Concurrency Processing, Tissue and Wound detection
TECHNOLOGY
OpenCV Python FastAPI three.js C/C++
Advanced Wound Care Platform
Wound healing is a complex regeneration process characterised by intercalating degradation and reassembly of connecting tissues and epidermal layer. Wound care issues lead to tremendous burden for patients and health care systems, with significant medical, social, psychological, and economic impact. We have built a platform using Deep Learning AI models designed to detect & measure anatomical and cellular structures, tissue segmentation, computer-aided disease diagnosis and prognosis. The products we develop are critical to keep our patients secure and make medical services more efficient
PROBLEM
Understand the user's behavior while they are using the apps to know how they are likely to purchase
SOLUTION
Built a big data analysis system to fetch user's daily data to analyze and predict the User Lifetime value
Big Data Analytics, Data Mining
TECHNOLOGY
Scikit-learn Python Flask BigQuery Pandas
User Lifetime Value 60 Prediction
As one of the largest network of user in the music game world, the problem of analysis, computation, and User Lifetime Value (User LTV) prediction are essential facets of the business planning methodology.
These concerns require them to invest time, money, and effort in user relationships to garner revenue and ensure the profitability of the entire enterprise.
User Lifetime Value (User LTV) refers to the approximate profit that can be gained over a well-defined period of time from a user or a group of users within an organization. The term ‘lifetime’ indicates that this value is measured for all transactions throughout the relationship with a user
We have deep dived ourselves in the big and massive amount of user behavior data to understand how user interacting with the app. From that insights, we find out the correlation between the user behaviors and the revenue they generate for the company.
This helps the company a lot in understanding user behaviors, user interests to design the app better, more focused on the paid users to generate more revenue.
PROBLEM
Detect the sound of notes played on a real piano to see if it matches with the sheet tutorial
SOLUTION
Built an optimize music note recognition model to listen and detect the sound from a real piano. The system returns the notes and chords that it recognize such as C, D#, Eb, F#m7, G, Am, B
Digital Signal Processing, Sound Sampling, Sound/Audio Recognition, Music Recognition, Speech Recognition, Spectrogram Features
TECHNOLOGY
Magenta Python Tensorflow
Piano Sound of Notes Recognition
Everyone loves music!
Piano is loveable to listen. However, not all/many people can afford to play piano. Moreover, piano has a steep learning curve for anybody in order to play it well and emotionally.
There are many ways to learn playing piano including the most traditional way which is enrolling to Conservatory of Music to pursue the professional path to become a pianist.
But again, not everyone can go that way. It takes too long (e.g 7 years to 11 years) to complete the program. There is a big portion of Piano lovers who would be happy as long as they can play Piano as amateurs.
We have developed an AI model to be able to recognize sound of notes played by real digital piano, acoustic piano, or keyboard.
The AI model is integrated into Piano teaching and tutorial apps in order to help Piano learners can play the real Piano according to the lessons.
Our work has contributed to create an educational and fun app for learning piano quickly and easily.
Research & Publications
We are rooted in academia with R&D advantages to pioneer in inventing and developing algorithms & machine learning models. These are some of our featured research and publications
Fine-grained Hand Gesture Recognition in Multi-viewpoint Hand Hygiene
This paper contributes a new high-quality dataset for hand gesture recognition in hand hygiene systems, named "MFH". Generally, current datasets are not focused on: (i) fine-grained actions; and (ii) data mismatch between different viewpoints, which are available under realistic settings. To address the aforementioned issues, the MFH dataset is proposed to contain a total of 731147 samples obtained by different camera views in 6 non-overlapping locations. Additionally, each sample belongs to one of seven steps introduced by the World Health Organization (WHO). As a minor contribution, inspired by advances in fine-grained image recognition and distribution adaptation, this paper recommends using the self-supervised learning method to handle these preceding problems. The extensive experiments on the benchmarking MFH dataset show that the introduced method yields competitive performance in both the Accuracy and the Macro F1-score. The code and the MFH dataset are available at this https URL.
Accepted for Oral presentation in The 2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC 2021)
What we help you
We are your expert partners for building impactful and innovative AI & Robotics products. This is a list of solutions and services we are opening to support you
Research and Develop AI & Robotics Products
Whether you are building an AI-based application or you are integrating AI into your current software such as eKYC, mining customer data, recommendation system, etc, these are our strength to help you.
We tackle these sub-problems in AI including Computer Vision and Pattern Recognition, Natural Language Processing, Digital Signal Image/Speech/Audio/Sound Processing.
Our background is mostly Math and Machine Learning, this is a solid foundation for us to tackle any problem and application in AI.
Web & App / IoT & Robot Development
Only developing AI models might not be enough for a usable and fully-featured product to functioning for end users.
You might need to wrap the AI core / models around by application platforms such as API, Web, Mobile, IoT, or Robots.
Or you have an idea and want to build a MVP to validate your business idea and market.
Our high-level and low-level software engineers are skillful and specialized in building up these application platforms to help you make impactful products.
Solution & System Optimization
If your current system, software, algorithm or AI model is not performing as you expected.
It could be low accuracy in face recognition of the AI model or the algorithm to predict the revenue is not accurate.
Our AI researchers sit down with you to brainstorm to find out the unoptimizied parts of the current system/algorithm/AI models.
We then research and devise a new and optimized solutions to make the system improve its performance.
Embedded Team / Staff Augmentation
You already have your own team taking care some parts of the product but your current team does not have enough capabilities for the full-product and full development speed.
Or your team wants to focus on their expertise and wants the other problems to be taken care by the other experts.
We will arrange our team as an embedded team to work and collaborate with your current team to both deliver every single part of the product as it best.
We're trusted by
Every day, we're trusted by entrepreneurs, funded startups, organizations, and large companies to design and develop AI & Robotics products. The people, leaders, managers, etc from these organizations have been in touched with us for collaboration and trusted us
Partnership
We are backed by our strong network of research insitutes, universities, and companies to research, solve challenging problems and push the limit of the current Articifial Intelligence
