saiwa services
saiwa is a B2B and B2C platform which provides artificial intelligence and machine learning software as a service (SaaS). Here, individuals and companies are empowered to explore and deploy AI services suited to their needs, with lower risk, without the need for expert knowledge of ML and with low initial investment. saiwa provides intuitive and easy-to-use web services for many AI applications. Users can try the services on their own data at no cost, forever! Below you will find a list of our current services. We are always open to work with clients to tailor our services according to their unique needs.
.
Advancing UAV-based EWC surveillance
To enhance Ducks Unlimited Canada (DUC)‘s capability for UAV-based surveillance of European Water Chestnut (EWC) through machine learning, we at Saiwa have previously implemented the initial version of the EWC detector software. In the second stage, we are in the process of finalizing the product’s features and upgrading its interfaces.
The two primary features to be incorporated in this stage are as follows:
1- Incremental learning for gradually training the deep network over time. This feature enables us to rectify false positive and false negative detections over time.
2- Reporting the 3D universal coordinates of EWC locations using drone configuration and temporal GPS data.
Aluminum Surface Defect Detection
We at Saiwa Inc. are pleased to announce that we have successfully completed an aluminium surface defect detection project in collaboration with AI-innovate Company for CASTechnology in Canada.
In this project, using machine learning techniques and the networks we provide in our Anomaly Detection service, we detect and localize the location of micro and macro defects on a casting line, including: crack, frost, frost patch, longitude frost and mold oscillation. This service is delivered via a simple user interface where users can run the defect detection APIs.
For more information about the project or if you find it interesting, please contact us at:Â support@saiwa.ai
European Water Chestnut Detection
European water chestnut (Trapa natans) is an invasive floating-leaved aquatic plant that is capable of out-competing native species and altering Ontario’s aquatic ecosystems. Saiwa inc. and DUC are working together to increase efficiency and performance of invasive species surveillance by using image processing and machine learning techniques.
Surface Defect Detection
To lower labor costs and enhance examination efficiency, more factories start to employ embedded machines for product quality control instead of manual examination in the manufacturing field. In the past several decades, automated surface inspection (ASI) have become more widespread as hardware and software technologies have developed rapidly. At saiwa we investigated several types of Surface Defects using 15 different datasets. We also provided a simple UI so that users may try the algorithms freely on their own images and in case of interest.
Corrective Exercise
Corrective exercise is a technique that leverages an understanding of anatomy, kinesiology, and biomechanics to address and fix movement compensations and imbalances to improve the overall quality of movement during workouts and in everyday life.
saiwa corrective exercise mobile application uses the state-of-the-art algorithms in pose estimation to detect a user’s pose (skeleton detection), then evaluate the vector geometry of the pose during exercise to provide useful feedback.
Contrast Enhancement
Contrast enhancement improves the perceptibility of objects in a scene by increasing the difference in brightness between objects and their backgrounds. Contrast enhancement of images has a wide range of applications such as medical imaging, industrial machine vision, and astronomical imaging. With the Saiwa Contrast Enhancement Demo, you can test a fast and robust local enhancement method called LLCC on your own images.
Image Restoration
Image restoration is the process of recovering an image from a degraded version. Image restoration is a critical step in many image processing applications such as medical imaging, remote sensing, security, and digital forensics. Saiwa offers three important restoration services: Denoising, Deblurring, and Deraining.
Resolution Enhancement
Resolution enhancement, or super-resolution, is the process of increasing the resolution of an image or video by generating missing high-frequency detail from low-resolution input. The goal is to produce an output image with a higher resolution than the input image while preserving the original content and structure. Image enhancement has many applications, including security and surveillance imaging, medical imaging, image generation, and satellite and astronomical imaging. In Saiwa’s resolution enhancement service, two CNN-based methods are provided: Residual Dense Network (RDN) and Residual in Residual Dense Network (RRDN).
Image Inpainting
Image inpainting is a technique that can reconstruct missing or damaged regions in an image using information from surrounding pixels. It can be used for various image processing tasks, such as removing unwanted objects, enhancing old photos, filling gaps, or creating artistic effects. Saiwa proposes a two-step generative method for deep image inpainting called DeepFill v2. This method fills large and multiple areas of an image without the usual boundary artifacts, distorted structures, and blurred textures inconsistent with the surrounding areas that we observe in other deep networks.
Object Detection
Object detection is a computer vision technique for locating instances of objects in images or videos. Object detection has a variety of applications, some of which are surveillance and security, traffic monitoring, video communication, robot vision, and animation. Saiwa’s object detection service uses three state-of-the-art deep neural networks: Detecron2, YOLOv5, and YOLOv7.
Pose Estimation
Pose estimation is a computer vision task where the goal is to determine the position and orientation of a person or animal. Typically, this is done by predicting the location of certain key points such as hands, head, elbows, etc. in the case of human pose estimation. Human pose estimation is used in a wide range of applications, including human-computer interaction, action recognition, motion capture, motion analysis, augmented reality, sports and fitness, and robotics. You can try both bottom-up and top-down pose estimation methods at Saiwa: OpenPose and MediPipe are the bottom-up and top-down deep networks, respectively.
Face Detection
Face detection is a computer technology used to find and identify human faces in digital images and video. Face detection technology is often used for real-time surveillance and tracking of people. At Saiwa, we provide two approaches to face detection: the Multitask Cascaded Convolutional Network (MTCNN) method and the Dlib open-source cross-platform library. The former is more accurate and takes advantage of new developments in deep learning, while the latter is known for its effectiveness and minimal computational complexity.
Count Objects
Computer vision object counting is the process of using intelligent algorithms to automatically detect and count the number of objects in an image or video stream. It has many real-world applications such as traffic flow monitoring, crowd estimation, and product counting. At Saiwa, we offer three deep learning methods in our Count Objects service: Detectron2, YOLOv5, and YOLOv7. All of these deep networks are new and robust object detection methods.
Anomaly Detection
Anomaly detection is about examining specific data points and detecting rare events that are suspicious because they’re different from the established pattern of behavior. Some areas where anomaly detection is popular include Manufacturing (aluminum, steel, or metal surface defect detection), Human Resource Management, Surveillance Systems, Medical Informatics (diagnosis, disorder detection), and Defect Detection. We have studied several types of surface defects and will add more and more anomalies in the future. Multiple classification and segmentation deep networks are used for each case and dataset.
Face Recognition
Facial recognition compares a human face captured from a digital image or video frame to an identity stored in a database of faces. AI-based facial recognition solutions help identify suspicious behavior, locate known criminals, and ensure safety in crowded places. At Saiwa, we provide a generic facial recognition interface to explore this powerful application online. We provide two face recognition methods with two different face detectors: Dlib and MTCNN.
Image Annotation
Image annotation refers to the process of annotating images from a dataset with labels to train a machine learning model. Annotation is a way to transfer human high-level knowledge of the image content to the model. Different models require different labeling: Boundary Annotation, Bounding-Box Annotation, and Classification Annotation
Deep Learning
Deep learning is a machine learning technique that teaches computers to do what humans do naturally: learn by example. Deep learning models can recognize complex patterns in images, text, sounds, and other data to produce accurate insights and predictions. Different types of deep learning models include: convolutional neural networks (CNNs), autoencoders, deep belief networks, recurrent neural networks. In Saiwa’s deep learning service, two CNNs are currently provided for training on user-specific image datasets (i.e. Detectron2 and Yolov5).
services
users
applications
Testimonials
Here are a few comments we received from experts and potential users.
Kyle Borrowman
Biologist at Ducks Unlimited Canada
Jean-Michel Morel
Centre Borelli, ENS Paris-Saclay
Kate Withers Hess
Innovation and Business Development Strategist
Partners
Blog
You can find all the blogs here
