With a diverse range of successfully completed projects, our portfolio showcases our expertise in various fields. We specialize in artificial intelligence (AI) and its subsets such as machine learning (ML), deep learning (DL), data science (DS), and OpenAI’s transformative technologies like ChatGPT. Furthermore, our proficiency extends to natural language processing (NLP), encompassing text-to-speech synthesis (TTS) and automatic speech recognition (ASR). Additionally, we excel in mobile app development, as well as delivering exceptional web design and development services.
Artificial Intelligence
AI Sign Language Recognition
An experimental sample for sign language recognition based on video recognition model. The goal is to create a universal model for gesture classification, which can later be used in translators and in various systems to convert gestures into text. To achieve this goal, frameworks, libraries and architecture such as Tensorflow, Pytorch, Keras, Pandas, OpenCV, CNN (VGG), LSTM, and VideoSwin Transformer were used. The result is the model capable of predicting words and phrases based on the isolated gestures provided.
Revolutionizing EU Real Estate Tech
Developed an advanced platform for property management and real estate marketplace analysis for EU-based agencies. This platform combines NLP (Transformers, BERT) and Computer Vision technologies to enhance property value estimates and risk assessment. Features include text-to-image search, image tagging, photo compliance checks, and watermark detection.
AI-Powered Virtual Teacher Bot
Virtual teacher application: Developed a Telegram bot for a virtual teacher using OpenAI technologies. This project involved advanced prompt engineering with OpenAI, incorporating the OpenAI API in Python, and integrating Text-to-Speech and Speech-to-Text functionalities.
For License Plate Recognition Camera Systems, researched and developed the detection, segmentation, and recognition (OCR) of license plates. Implemented duplicate searches and consolidated results from multiple recognition algorithms.
For License Plate Recognition Camera Systems, researched and developed the detection, segmentation, and recognition (OCR) of license plates. Implemented duplicate searches and consolidated results from multiple recognition algorithms.
Multilingual NLP Chatbot Development
NLP and chatbot development aim to bridge the gap between human language and computer understanding, enabling more intuitive and intelligent interactions between users and machines. Developed and integrated a multilingual user support chatbot leveraging NLP technologies. Technologies used include Python, PyTorch, spaCy, and capabilities such as text classification, sentiment analysis, and language detection.
Cloud-Integrated IoT Vision System
A cloud-based Situational Awareness System was successfully deployed into production, showcasing extensive practical experience in integrating IoT and Computer Vision technologies. Utilizing AWS Greengrass for IoT, team incorporated advanced object detection, image recognition, and image segmentation capabilities. This system significantly enhances situational awareness by processing and analyzing visual data in real-time, providing valuable insights and improving decision-making processes. The work involved setting up and configuring the cloud infrastructure, developing and integrating the necessary software components, and ensuring seamless operation and scalability of the system.
Advanced License Plate Recognition
For License Plate Recognition Camera Systems, there were research and development focused on detecting, segmenting, and recognizing license plates using Optical Character Recognition (OCR). Additionally, it was implemented duplicate search functionality and consolidated results from multiple recognition systems.
Innovative Neural Network for Facial Analysis
A novel neural network architecture and devised a methodology for its training were developed, accompanied by the creation of a specialized dataset tailored to the task. The algorithm is designed to perform simultaneous estimation of facial image quality and weighted aggregation of face embeddings derived from multiple photographs using image inputs.
Precision Vehicle Speed Monitoring
The implementation and deployment of a Vehicle Speed Estimation system in a designated district of the city was undertooked. This project utilized advanced image processing techniques coupled with object detection algorithms to accurately calculate the average speed of vehicles. The primary objective was to enforce adherence to a speed limit of 20 km/h within the monitored area.
Advanced Check Amount Recognition System
The extraction of significant features was a crucial step, as it allowed us to identify the unique characteristics of the handwritten and printed amounts. This process required detailed data analysis to ensure that the extracted features were both accurate and useful for training our models.
Training neural networks was a central part of the project. The project is focused on creating and refining models that could reliably recognize and interpret the amounts on the checks. This involved extensive experimentation and tuning to achieve the desired levels of accuracy and reliability.
Overall, this project showcased ability to handle complex image processing tasks, from initial data cleaning to advanced neural network training, culminating in a robust solution for check amount recognition.
Training neural networks was a central part of the project. The project is focused on creating and refining models that could reliably recognize and interpret the amounts on the checks. This involved extensive experimentation and tuning to achieve the desired levels of accuracy and reliability.
Overall, this project showcased ability to handle complex image processing tasks, from initial data cleaning to advanced neural network training, culminating in a robust solution for check amount recognition.
Enhancing Visual Clarity with OpenCV
The project is focused on leveraging image processing and computer vision techniques using OpenCV and Python. The initial phase involved implementing a range of algorithms tailored to image manipulation tasks. Subsequently, the team delved into developing a specialized deep neural network using the Keras library. This neural network was meticulously crafted to achieve precise background removal from video footage, aiming to enhance the visual quality and clarity of the captured content.
