Category: AI Portfolio

Face your attendance! | AI Portfolio

Domain: Human Resource

Face recognition technology is not new to many of us. We have seen all of it in many science fiction movies where secret doors are unlocked only after a face scan. The journey from sci-fi to reality has not been any different. Today we unlock our phones using the same age-old technology. While the usage of this technology has proliferated now, it has has been around since the 1960s. It is only the recent developments and needs that has led to its massive success and growth. Not limited to the phone unlocks, face recognition technology is now being used as a path-breaking tool in many of the other industries. It is being employed for a wide range of use cases – preventing crime, protecting events, making travel more convenient, marking attendance, etc.

Face recognition technology

Problem Statement: 

Building a highly efficient and robust attendance system using face recognition

Technology and Situation Overview: 

In the recent past, automatic face recognition technologies have changed the way we function by bringing in many improvements. Today, a human face is not just meant to be for social interaction but is also extensively utilised by devices to perform certain actions.

At a corporate or a business level, smart attendance using real-time face recognition is a real-world solution which comes with day-in-day-out activities of handling manpower at work. This simply means that it is a process to recognise the employee face for marking attendance by using face biometrics based on high-definition monitor video and other information and data technologies. The technology follows three steps in order to mark an attendance – face detection, face capture, and face match! There is no need for one to punch-in or punch-out, capturing face recognition app does it all.

Why should we solve this problem? 

Decreasing the turn-around-time to capture attendance. The fastest and most accurate way to capture attendance real-time

How did we solve this problem? 

Our solution includes two parts:

  1. Face Detection: We used a pre-trained MTCNN model to detect the faces.
  2. Face Recognition: We used a pre-trained FaceNet model trained on VGGFace2 to recognize the faces.

Inference Pipeline on a Live Video Stream:

We have made a web application which consists of two parts :

  1. Registration: This registration page is made to register new onboarding people in the organization. So when a new person is detected his face is passed through our MTCNN model and then the detected face is passed through our Facenet Model to generate the face embeddings for the newly onboarded person. This generated embedding list is then maintained in an Annoy CSV file.
  2. Attendance: Through the attendance page we continuously send a live video feed to our backend code. Where each frame of the video is passed through the MTCNN model to detect the faces. And then these detected faces are then passed to our FaceNet model for recognising those faces and marking attendance.

What technologies did we use?

  • Python
  • Pytorch
  • cv2
  • PIL
  • numpy
  • pandas
  • Flask
  • js
  • Google Cloud.

What was the business Impact:

  • Developed using state-of-art face detection and recognition models, it has an accuracy measure of 99.63%
  • Error-free attendance marking as no manual intervention is required
  • No need for specialized hardware for installing the system in the office
  • Contactless experience


Unsolicited Recommendations: Tool to keep audience engaged

Domain: E-Commerce

In recent years, we all have seen an explosion in the ways that consumer e-commerce brands function. Most of them are using big data and AI for their advantage by leveraging user experience. Alexa, Siri, chatbots, drip-emails, and category recommendations are just some examples of how the brands have changed the buying patterns and behaviours. It would not be wrong to call it the tip of an AI ice-berg. This is how Artificial Intelligence development is bringing in the change in our everyday lifestyle.


E-commerce future technology

Today, the development of artificial intelligence has made the world a profit-driven global enterprise where sales need to happen regardless of what time of the day it is. Businesses need to be available to the customers at all hours of the day. In the sector of e-commerce, AI development can be helpful for the brands to achieve this objective by optimising their e-commerce framework. It helps companies to gather and investigate data in real-time, thereby facilitating more efficiency and competence in the business. This, in turn, has led to creating a more personalised experience by identifying different patterns of consumer behaviour.

Staying ahead of the time, and acing at what we do, here is the story of how we incorporated AI and machine learning application in an e-commerce framework.

Problem Statement:

Improve ‘In-Category’ and ‘Cross-Category’ product recommendation on E-Commerce websites across all categories using Artificial Intelligence


In-Category: Similar set of items (degree of similarity may vary)

Cross-Category: Items that complement each other

Technology and Situation Overview:

A recommendation is a type of information filtering system that uses AI algorithms to provide the most relevant items, and/or content to a user. These engines can incorporate a variety of data, including user purchase behaviour, user browsing behaviour, user demographics, and real-time triggers.

With the dearth of goods available for one to choose, customers usually tend to lose out on goods they have previously viewed or liked. In-app recommendations enable them to be spoilt for choice. Built with customer behaviour analysis at its core, the recommendations can keep them hooked enabling to view products that they might be interested in.

Why should we solve this problem?

Recommendation engines can help e-commerce players increase engagement, lift conversion rates, and improve revenue.

How did we solve this problem?

For In-Category recommendations, we used the client’s inventory data-set. Upon mapping and fragmenting the styles we developed a model using the below approach:

Trained a ResNet50 architecture network on Client’s Inventory Dataset using transfer learning. Generated cloth embeddings by doing a forward pass on trained ResNet50 keras and collected the embedding from the preceding connected layer using our client’s inventory Dataset, and later stored in an annoy file.

During real-time In-Category recommendation, the application took the image captured by the customer and passed it through the Resnet50 keras and collected the embedding for the same. Generated the In-Category recommendation using nearest cosine similarity measures over the Annoy Embedding file.

For Cross-Category recommendations, we collaborated with the client’s fashion Stylist to generate the fashion-sense dataset, and the model was build using the below approach:

Trained a multi-layered Bi-Lstm RNN network using the embedding file of the fashion-sense dataset to make the network learn about fashion sense.

During the real-time Cross-Category recommendation, the application took image captured by the customer and passed it through the Bi-Lstm RNN and generated the embedding of products of various categories which are related by networks fashion sense.

What technologies did we use?

  • Python
  • Matplotlib
  • Pytorch
  • Flask
  • Cv2
  • Google Cloud

What was the business Impact:

  • Enhanced Product Recommend System using AI development improved In-Category and Cross-Category Sales
  • Exclusive brand websites can enjoy the recommendation solutions which big e-commerce players regularly use
  • Algorithms are applicable for both web and mobile
  • Creating a brand-persona by acting as a virtual stylist by recommending the best fashion

How can Artificial Intelligence enhance Travel?

#Wanderlust is trending on social media channels. There is no denying that curated travel experiences have been at the heart of the travel industry.

People, these days, participate actively to enhance their travel experiences. As the travel industry is picking numbers, the demand for personalization is growing.

This sudden rise in customization demands has necessitated the use of forward-technologies to cover the gap. As it appears, Artificial Intelligence enhances travel experience to fulfill the gap of demand and supply.

The emergence of Artificial intelligence to enhance Travel!

Technologies like Artificial Intelligence (AI) and Machine Learning assist travel enthusiasts to a great extent. They help people get a faster, safer, and more personalized travel experience.

As we speak, Artificial Intelligence is already making a significant impact on the travel and tourism industry. The role of Artificial Intelligence in the functions of traditional human cognition has made life easier.

For giant business houses with frequent travel requirements, AI proves to be beneficial in many ways-

  • Saving time and money.
  • Provide memorable travel experiences.
  • It ensures that business travelers reach their desired destinations without trouble.

In the midst of advancement, it is very well accounted that the expectations of travelers have changed over time. Factors like price, comfort, and services on offer are more crucial than anything else. These expectations can be met successfully by using machine learning.

Artificial Intelligence is a term synonymous today with Internet businesses, e-visa processing, citizen services, travel hubs, and contact centers. The technology has entirely changed the general operational activities in a brief time.

Moreover, speech recognition, language translation, and visual perception are impacts felt in all businesses these days. When AI was not renowned as it is today, experienced industrialists around the world expected disruption by the use of AI.

This particular fear of disruption has not been too important, though. Major industry players are aware of the use of machine learning to provide quick services to clients.

how can Artificial Intelligence Enhance Travel

What exactly does AI do for travel

When it comes to tourism, there are several activities where Artificial Intelligence enhances travels and tourism experience.


Operational windows of contact centers play a crucial role to address visa related queries without too much trouble. AI creates a smooth link between business and visa applicants.

From selecting a preferred language or evaluating the wait time, AI chatbots take in little details to optimize contact centers. Furthermore, getting a customer’s automatic identification or any other piece of relevant data becomes a great experience through a simple AI algorithm.

Data Security:-

With new trends coming up in technology, the importance of data security and privacy has increased immensely. To get a global visa, many outsourced companies need to follow the guidelines of data security and embassy protocol.

Keeping cyber threats into consideration, Visa providers have become more watchful. They now pay more attention to security audits, data encryption, and make use of fool-proof password control systems.

Security here is the first priority and should complement the pace of innovation in the tourism sector.

How AI Is Redefining Travel Today

“Relevance” is a keyword and an important winning factor with AI in the travel sector.

It boils down to the point whether AI has enough potential to bring about a significant change in the ways travel experiences are delivered to customers.

There are a few key areas where the use of Artificial Intelligence can have a lot of impacts to give a better experience for the customers-


In today’s time, a traveler has access to every imaginable piece of information on a single website. Through these websites, travelers plan their destinations and compare different options. Budgets, bookings, and their cancellations come after that.

All of the above activities involve a reading of different descriptions, terms, and instructions before people arrive at a decision. A great alternative is to provide other apps with the role of reducing the interaction.

Bots using Natural Language Processing (NLP) perform the task of nailing more personal interaction for AI through context. The bot finds it easier to understand the meaning of a customer’s query and sort it.

Together, NLP and AI add a lot of weight to every travel-related activity. The immense scalability of these bots is also a fantastic facet to their attributes. AI assistance will go a long way to reduce the inconvenience faced by frequent travelers.


Travel requires constant monitoring of related documents by different sets of people. There are complex on-boarding and off-boarding processes involved. Facial recognition is a positive way to bring an end to these tiring paper-driven procedures.

Facial recognition will allow travelers to quickly move through airports, customs, immigration, and board flights without having their documents validated.

When combined with blockchain, customers find it easier to visit restaurants and duty-free stores. They can entertain themselves after just a face scan. Blockchain technology ensures that reliable data is always available to complete transactions.


Airports and airlines have begun to replicate shopping malls and huge retail outlets. Everything from blankets, seats, and hotel rooms are sold there. Machine learning has emerged as a new trend to assist with the sales procedure.

By the use of big data and machine learning, airlines can develop recommendation engines. The role of these engines is to personalize the offers around products from their catalogs.

AI and machine learning make personalization easy. This is because travelers expect travel companies to understand their preferences and offer better deals.  Machine learning also uses external data to help travelers to make quick decisions.


The applications mentioned above of Artificial Intelligence, or AI, have one thing in common; they lead to time reduction and also improve the accuracy of outcomes generated.

In an industry where time is of the essence, and data is fluctuating, AI provides many capabilities to ensure things work out correctly.

Did you like this post? Have an actionable idea on AI? Hire us and open the doors of wonders. Let’s work together. Contact us!




Image Scanning and Processing with ML Models

Image Scanning & Processing with Machine Learning models

One of our Fortune 500 clients in the logistics industry wanted to extract various product-related information by scanning images through a machine learning model. This scanned information had to then be supplied to a custom web application for further utilization and analysis.

Image scanning for logistics

The Objective

The image scanning and detection had to happen on the below aspects

  • Identifying the object in the image
  • Localization of the object
  • Measuring the width and height of the objects in the image


The GoodWorkLabs Machine Learning Solution:

Our data scientists used the Faster-RCNN algorithm to solve the problem statement. We followed the below procedure to achieve the desired results.

  • We ran the image through a CNN to get a Feature Map, a matrix representation of the image between a neural network layer
  • We ran the activation map through a separate network called the Region Proposal Network(RPN), which identified the bounding boxes (interesting regions) for those objects. This output (regions) was then passed on to the next stage.
  • Each and every output of the bounding boxes was analyzed and the most appropriate bounding box coordinates was accepted.

Faster-RCNN works quicker because we pass the activation map through a few more layers to find the bounding box (interesting regions). This forward pass continuously takes place and during this training phase, the ML model continues to learn. Errors (if any) are captured at this stage and with continuous learning, the model becomes efficient in predicting the classes and bounding box coordinates.

For calculating the height and width of each object we continued to iterate every object in the image and calculated values using OpenCV.

Faster Rcnn - ML model

Image reference:



To perform this image scanning process, we had a well-annotated object in each of the images in the dataset. We had around 1000 labels for each object.


How did we train our ML Model:

  • We downloaded pre-trained models and weights. The current code support is VGG16 
  • We also got access to pre-trained models which were provided by pytorch-vgg 
  • In the next step, we trained our model from fine-tuning to a pre-trained Faster R-CNN model. We followed this approach because a pre-trained Faster R-CNN contains a lot of good lower level features, which can be used generally.
  • We trained the model for 150 epochs.


GPU utilization:

The models were then exported to Microsoft Azure’s GPU for better performance. The expected inference time for a given image is ~0.2 seconds.


Technology Stack:

The technology stack used to implement this image scanning ML model was Python, Pytorch, OpenCV, Microsoft Azure.


The GoodWorkLabs AI and ML solution:

Are you looking for a partner who can build advanced AI/ML technologies for your business and make every interaction of your business intelligent? You are at the right place.

We love data and we are problem solvers. Our expert team of data scientists dives deep into solving and automating complex business problems. From Automobile to Fintech, Logistics, Retail, and Healthcare, GoodWorkLabs can help you build a custom solution catered for your business.

Leave us a short message with your requirements.




Travel Recommendation App using AI & ML models

High-performing Travel Recommendation Engine built with AI/ML models

One of our Fortune 500 clients had a community-based travel app that helped create trips for its users. Through this app, users could explore the community, take trips to nearby places, and also browse through their previous trips in the travel history.


Travel App - Artificial Intelligence in Travel


Our data scientists at GoodWorkLabs were entrusted with the task to make the above mentioned mobile app engaging, intelligent, and personalized. We had to create recommendation systems as an advanced feature by using Machine Learning models.

We realized that recommendations could be made to users based on nearby attractions, restaurants, hotels, etc. The nature of these recommendations had to be as below:

  • Users will be recommended with places they would like to visit based on their previous travel history.
  • Users will be recommended with nearby tourist attractions when they visit a particular place.
  • Users will be recommended places based on their preferences and tastes.
  • Users will also receive recommendations from similar travelers who share the same interests.


Recommendations by using Machine Learning models

To build an effective recommendation system, we trained the algorithm to analyze key data points as below:

  • On-boarding information: To capture user data at the sign-up stage of the web application
  • User profile: To suggest recommendations by analyzing data from the user’s previous visits on the profile 
  • Popularity: To suggest recommendations based on user ratings that were collected in the form of reviews
  • Like minds: To analyze data and match it against the likes of different users and populate recommendations accordingly.


App screens that populated ML recommendations

We programmed specific screens on the mobile app to display the recommendations. Below were the mobile screens on which ML recommendations were displayed:

  1. Attractions
  2. Trips
  3. Restaurants
  4. Nearby Cities
  5. Ad-hoc Plans
  6. Search (when users search for places to visit)


Types of Recommendation systems:

1. Content-Based recommendation

Based on the details keyed in by the user at the signup stage and in the whole travel process, the content based recommendation system analyzed each item and user profile. All this data was stored and the system was optimized for continuous and smart learning.

2. Collaborative filtering/ recommendation

In this recommendation system, the system looked for similar data inputs keyed in by different users. This was then continuously compared against other data. Whenever there was a match, the system recorded the instance and populated a set of recommendations that were common to that set. In this recommendation system, user interactions played an important role.

At GoodWorkLabs, we suggested a hybrid model of both the above-mentioned approaches for optimal performance of the recommendation system.

Tech Stack:

The tech stack we implemented in building these Machine Learning models were Python, Tensorflow, Sklearn, iOS CoreML, Elasticsearch.


The GoodWorkLabs AI and ML solution:

Are you looking for a partner who can build advanced AI/ML technologies for your business and make every interaction of your business intelligent? You are at the right place.

We love data and we are problem solvers. Our expert team of data scientists dives deep into solving and automating complex business problems. From Automobile to Fintech, Logistics, Retail, and Healthcare, GoodWorkLabs can help you build a custom solution catered for your business.

Leave us a short message with your requirements.

Artificial Intelligence (AI) Racing Assistant

AI Racing Assistant to Enhance Driver Experience

One of our Fortune 500 clients in the automobile industry wanted to analyze and improve their racetrack experience. The racing car is equipped with more than 100 sensors and these were programmed to capture all activities of the car such as steering wheel angle, acceleration, engine running state, etc. 


AI solutions in Automobile industry

The Objective:

The GoodWorkLabs AI team was given the task to identify the optimal path of the vehicle by analyzing the complete racing track with other tracks and also the overall racing experience.


AI/ML Implementation & Solutions:

We first analyzed the track using sensor data and then implemented state-of-the-art Deep Q-learning with Tensorflow.

Our Deep Q Neural Network took a stack of n frames as input. These pass through the network, and output a vector of Q-values for each action possible in the given state. We then take the biggest Q-value of this vector to find our best action.

In the beginning, the agent does not perform well. But with time and continuous learning, it began to associate frames (states) with best actions. Pre-processing was a very important step as we wanted to reduce the complexity of our states and reduce the computation time needed for training.

For that to happen, we greyscale each of our states. Color does not add important information (in our case, we just had to find the optimal path). This is an important saving since we reduced our three color channels (RGB) to 1 (greyscale).


Tech Stack: 

The tech stack used to develop this model was Python, PyTorch.

Below are some visualizations of the optimal path identified by our AI model.

UX Designs for AI and ML


The GoodWorkLabs Artificial Intelligence and Machine Learning solution:

Are you looking for a partner who can build advanced AI/ML technologies for your business and make every interaction of your business intelligent? You are at the right place.

We love data and we are problem solvers. Our expert team of data scientists dives deep into solving and automating complex business problems. From Automobile to Fintech, Logistics, Retail, and Healthcare, GoodWorkLabs can help you build a customized solution to cater to your business.

Leave us a short message with your requirements.



Ready to start building your next technology project?