Vision Engineering Lab

The number of theft activities in banks, jewellery shops is constantly increasing despite the fact that all these places are equipped with CCTV cameras 24x7. These cameras fail to detect human intrusion instantly at the time of theft and give information only after the entire event. AI Janitor is AI powered intelligent assistant that uses deep learning based computer vision algorithms to examine and check every single frame in CCTV footages at real time instantly to identify any human intrusion in highly secure zones and once human is identified immediately the system makes a phone call to the corresponding owners along with alarm and captured images of it thefts.

Generally, most of the information communicated in our day to day environment is of visual form often in form of pictures, sign boards, name plates, etc. This visual form of information communication is barely accessible by most of the visually impaired people and they often depend on external human help to recognize and identify these. Things get even worse when a blind people are navigating to a new environment where they face numerous uncertainties. In this work, we are solving this problem, using DL based computer vision algorithms called Prakash - scene text recognition. The solution speaks out the textual content present in real environment from an image that is captured by a Smartphone.

Human race has overcome numerous pandemic and epidemics like Spanish flu, SARS, cholera, plague, etc since ages and COVID 19 pandemic is one among them. COVID 19 being the recent one is much different than the others due to the contribution of AI in diagnosis and prediction of COVID 19 patients. Among the various use cases, one widely used area is medical diagnosis. AI and deep learning based algorithms are exploited enormously by data scientist to support radiologist in early prediction and detection of corona patients. Subsequently, in this work, we utilize wavelet based Convolutional Neural Networks for detecting and recognizing of COVID 19 cases from chest X ray images. Currently, previous works utilize existing CNN architectures for classification of healthy and affected chest X rays, however these networks process the image in a single resolution and loss the potential features present in other resolutions of the input image. Wavelets are known to decompose the image into different spatial resolutions based on the high pass and low pass frequency components and extract valuable features from the affected portion of lung X ray efficiently. Henceforth, in this article, we utilize a hybrid CNN model of wavelet and CNN to diagnose the lung X rays. The proposed CNN model is trained and validated on open source COVID 19 chest X ray images and performs better than existing state of the art CNN models with an accuracy of 99.25%, ROC-AUC value of 1.00 and very less false negative values. Further, the performance of our model is validated by Gradient Class Activation Map visualization technique. The visualization of feature maps clearly indicates that our proposed network has perfectly extracted features from the corona virus affected portion of the lung.

Nowadays many organizations are shut down due to the pandemic, so we created a project where we trained a model which will do two things to detect mask and also recognize persons, so if someone doesn't wear a mask he will be recognized and an alert message will be sent to him as well as the higher officials. Everyone knows 'Face Mask Detection', but why can't they see that implemented elsewhere? Here we implemented end to end working model of Face mask detection which can be used in organizations, colleges, schools. They could also view a dashboard where they could access all the CCTV in their organization; for the officials to monitor. This can make all us work or study again in any organization. It not only detects the mask on the face instead we designed the ML model as, it searches every face in the frame with the database and verify it and alerts the particular employee/student/person and the higher officials with an alert message, and displays all these in a dashboard. This can almost change the current scenarios in society.

Alzheimer's disease (AD), is a chronic neurodegenerative disease that usually develops gradually over time. It is the cause of 60-70% dementia cases. However, identifying differences between Alzheimer's brain data and healthy brain data in older adults (age> 75) is challenging due to very similar brain patterns and image strength. Therefore, early detection of computer-assisted systems is a matter of great concern and in-depth research by researchers. In this work we have done multilevel classification of Alzheimer's disease using deep learning technique and bring forth the main brain affected regions which help the doctors to learn the regions pertaining to early causes of the disease.

Customer analytics and determining the exact number of people in commercial locations is gaining high importance nowadays.Especially in crowded environments like malls, shopping complex, hotels, airports, railway stations, etc getting the exact count of people is extremely crucial for security and surveillance requirements. Currently, human efforts are widely used in obtaining the exact count of people in an area. Therefore, to solve this problem, we utilize deep learning based computer vision algorithms to exactly determine the number of people within a campus from the video frames obtained from already installed CCTV cameras and deliver insightful business analytics for firms.

Inspection of brake components is very essential to detect the damaged manufactured parts before it is assembled in any vehicle. Manual inspection of brakes is extremely difficult since most of defects are very minute and cannot be identified by human eyes. Therefore, automatic inspection of manufactured brakes is indispensible to prevent failure of brakes and accidents. Previously, various research articles perform inspection of brake through conventional image processing and traditional image processing algorithms. However, these techniques are capable of identifying a single fault only and are less robust to detecting numerous faults. Further, the existing techniques hardly localize the exact location of faults in the surface of brake. In order to over these drawbacks, in this research we utilize deep learning object detection algorithms namely Single Shot Detector and Faster RCNN to identify and localize the exact location of fault on the brake surface. Furthermore, the proposed system is capable to detect different types of faults in a single algorithm and is robust to brake’s material surface, environmental and lightening factors. The deep learning algorithms are trained using transfer learning on custom collected dataset. The proposed algorithms deliver an accuracy of 95.64% and mAP of 73.2% on cylindrical grey shade brakes.

It is implemented in real-time and was developed for use in surveillance and security applications. This method identifies camera tampering by detecting large differences between older frames of video and more recent frames. A buffer of incoming video frames is kept and three different measures of image dissimilarity are used to compare the frames. After normalization, a set of conditions is tested to decide if camera tampering has occurred. The effects of adjusting the internal parameters of the algorithm are examined. The performance of this method is shown to be extremely favorable in real-world settings

With the emergence of the concept of “safe city”, security construction has gradually been valued by various cities, and video surveillance technology has also been continuously developed and applied. However, as the functional requirements of actual applications become more and more diverse, video surveillance systems also need to be more intelligent. Aiming at the problem of abnormal behavior detection, especially the low efficiency and low accuracy of brute force detection, a brute force detection method based on the combination of convolutional neural network and trajectory is proposed. This method uses artificial features and depth features to extract the spatiotemporal features of the video through a convolutional neural network and combines them with the trajectory features. In view of the problem that face images in surveillance video cannot be accurately recognized due to low resolution, two models are proposed: the multi-foot input CNN model and the SPP-based CNN model.