People working in banks face many problems, including approving loans. The project proposes a loan sanctions system that determines whether a loan should be provided to someone based on certain attributes. Although the bank follows strict rules and regulations, and conducts detailed background checks when approving loans, and keeps in mind the individuals ability to repay the loan, in this case, people often fail to repay the loan and have already given it to him. In this project, the system , propose for bankers will help them predict the trusted customers who have applied for loans, thereby increasing the chance of timely repayment of loans. This prediction is done using LSTM neural network. To improve the performance of LSTM , Whale optimization algorithm is used in this project.

Video tagging is a complex problem that combines single-image feature extraction with arbitrarily long sequence understanding. By improving at the task of tagging videos with useful metadata labels, we necessarily improve our ability to understand the content and context of video data. A lot of existing methods fail to precisely tag videos because of their lack of ability to capture the video context. The context in a video represents the interactions of objects in a scene and their overall meaning. In this work, we propose a novel approach that integrates the video scene ontology with CNN based YOLO algorithm for improved video tagging.

Reinforced concrete is the one of the versatile and economical material which is used in the world wide. Steel is used as reinforcement material in the concrete. Reinforcement embedded in the brittle concrete imparts tensile strength. Corrosion of these steel is a main factor that affects both durability and strength when subjected to the sever environment conditions. The concrete who’s PH below 10 will destroy passivity of reinforcing steel bars and leads for corrosion. This paper includes the experimental analysis on prevention of steel corrosion in concrete by using natural inhibitor allium sativum. As corrosion is longer process solution of HCL and water is used to increase rate of corrosion in in varying proportion 30% and 50% of total weight of solution. Artificial inhibitor epoxy resin is used for result comparison. Results of weight loss test and electrical resistance were showed that allium sativum is a good inhibitor and having inhibition efficiency between 90%- 95%.

In the construction Industry, the utmost utilized commodity is Concrete. Concrete has excessive strength and stability. While construction, huge amount of waste is generated. This waste can be used as admixtures, or can form a complete different product altogether. By excessive and mass production of these waste products, a hazardous effect can transpire on the environment. Hence we need to observe keenly and perceive a way to reduce or reuse this waste to create something prime and non-perilous for the environment. The all-inclusive performance of Reinforced Concrete Composite material is pretentious i.e affected than the individual performance of the concrete itself, due to which a new material was explored. This paper reviews that how concrete can behaves when cement can by replaced by various products like the GGBS (Ground granulated blast furnace slag), foundry slag fly ash etc.in different proportions some time these materials were used at an replacement level of 50%, which has lower heat of hydration, higher durability when compared with normal Ordinary concrete. Many a times used at a replacement level of 50%, which has lower heat of hydration, higher durability when compared with normal Ordinary concrete. This paper will reviews different properties of concrete when Alccofine material of used as cementitious material. Results were indicates that use of alccofine powder will enhance different properties of concrete.

Big data is a set of data that has images, videos, audio and many more. This unstructured information and unconventional qualities from conventional data sets are commonly connected with additional difficulties in putting away, breaking down and applying different systems or extricating results. Big Data examination involves reviewing huge measures of detailed information to discover modest examples or perceiving stowed away relationships. Big Data applications have been rising during the last years, and scientists from many systems know about the upsides of information extraction from this sort of issue. Notwithstanding, it cant uphold conventional learning approaches because of adaptability issues. A small collection of experiments has been directed on imbalanced information grouping for Big Data as a new regulation. The concern behind this is the challenges in adjusting standard procedures to the Map-Reduce programming style.

Friction stir welding (FSW) is a relatively modern process which is considered a suitable permanent join method for different industrial applications such as aerospace and automotive. Moreover, friction stir welding is preferred in contrast with other welding methods due to its ability to join similar and dissimilar metals with high efficiency. The current study presents a review for new researches taking in the consideration FSW process variables like welding speed, tool rotational speed and pin geometry for different metals alloys in addition to failure modes occurred in these alloys.

Optical character recognition (OCR) of Devanagari script characters has been a popular research topic for many years and still a lot of work is yet to be done. The strive to achieve utmost accuracy in recognizing the characters is ever-increasing and a task yet to be accomplished. The complexity of this particular script having very peculiar features has attracted many researchers to contribute in this field and showcase their work. Building an OCR for any script involves various phases, the two of vital importance being segmentation and recognition. In this paper, we will review each of the two in detail, by observing the existing methodologies used and their probable scope of improvement.

For the past few years, neural networks are maturing, and the application domain for the neural network is increasing. Due to the rise in unstructured Image-based data, theres a need for understanding data based upon visual features, not on textual data. Image Captioning in Deep learning is the process to understand different objects in the image, try to build a relation between those objects, and give a sentence that is semantically and syntactically correct. To serve this purpose, Image Captioning uses the combined architecture of Computer Vision and NLP-based Networks as Encoder-Decoder Architecture. In this Survey paper, we discuss the paper[1], which proposed to use Efficient CNN-LSTM based Network using Network Compression. The Network uses CNN-based VGG-16 as encoder and LSTM as decoder network. Network techniques are used such as Quantization and Pruning, to reduce model size up to 73.1% and reduce inference time up to 71.3% and increase BLEU score to 7.7% as compared to uncompressed network.

Drowsiness is one of the main contributing factor in many traffic accidents due to the clear decline in the attention and recognition of danger drivers, diminishing vehicle-handling abilities. A new approach towards automobile safety and security in an autonomous area is primarily expected on the automotive system. To overcome this problem, here is a solution namely the driver drowsiness alert system, which gives an alert by watching each driver’s eye movements in real – time environment. In a process of detecting a person falling asleep, a real time eye detection of the person is acquired. The image processing of an eye through mood detection and adjacent to count of heart rate is analyzed over time to identify drowsiness and fatigue situations, and characteristics indicative of the person falling dormant are determined. To determine openings and closings of the eyes, eye aspect ratio is analyzed to determine the width and height of the eyes, mood discernment analysis through Keras algorithm and heart rate diagnosis manoeuvring heart beat sensor

The ultimate aim of agriculture or farming in India is not only limited to growing of crops but is also associated with the economic growth of farmers and labours. Rice is one of the staple food crop of our country. Basically in India establishment of rice depends on the availability of moisture, climatic condition, age of the variety, availability of inputs & human labour. Mechanization in agricultural sector is advancing in developing countries like India. Rice is a labour-intensive crop and requires about 80-90 labour days per acre. Timely availability of labour and water for various activities of rice is becoming a problem. Hence to overcome these issues there is a need of mechanization in the field of rice cultivation by using rice transplanter as major tool in this process. There is also need for designing and developing an economical and user-friendly rice transplanter for small scale farmers in order to increase the production as well as the quality of rice. In this paper manual rice planting machine along with their merits and demerits has been discussed by studying various aspects of transplantation related to rice and its field performance which are beneficial to the society and farmers. A rice transplanter is specialized equipment best fitted to transplant rice seedlings on the wet muddy paddy field. This paper is focused on developing a machine which addresses labour problems faced by small scale farmers. The newly developed rice planting machine, can harvest upto two rows of paddy at a time. Outcomes are Reduction in manpower by 80%., Reduction in time by 50%, Reduction in production cost by 50%, Increase in product quality.

Heart rate is the number of heartbeats per minute which is the number of contractions of the ventricles (the lower chambers of the heart). For some human beings it might also be too excessive (tachycardia) or too low (bradycardia) which can cause clinical problems. In the current pandemic situation estimating or monitoring heart rate of a subject with the help of instruments can be risky as it involves physical contact and also one has to visit the clinic which might not be possible all the time. Research focused on non-contact based systems has increased over the past years. Existing systems include those which require contact, some having restrictions on skin tone, and others often involve high costs and complex application of sensors. This paper focuses on real time monitoring of Heart rate of multiple people simultaneously. It is obtained through a real time video using a webcam of laptop/computer. As blood circulation causes facial skin variation, therefore facial video is considered. Signal processing methodssuch as Fast Fourier Transform have been applied on the region extracted after applying colour magnification channels in video recordings. Application also sends a notification via a text message to the specified contact number if the Heart rate is below or above a given range. This algorithm is easy to implement, low cost and simple for real time application. Testing included trials to calculate heart rate of multiple people simultaneously. Once such a trial involved detection of faces of four people and their Heart rate calculation at the same time. Measured value of Heart rate in comparison with the ideal method was within justifiable range. Further, work can be done on environmental conditions which can be very useful in many real time applications.

We propose in this paper a fully automated deep learning model, which learns to classify human actions. Human action recognition is an important application domain in computer vision. Its primary aim is to accurately describe human actions and their interactions from a previously unseen data sequence acquired by sensors. The ability to recognize, understand and predict complex human actions enables the construction of many important applications such as intelligent surveillance systems, human-computer interfaces, health care, security and military applications. Human activity recognition (HAR) is the outcome of a similar motive. Recent times have seen the theory of Human Activity Recognition (HAR) catering to multiple challenging applications built on the increase in ubiquitous, persuasive computing.