Search

Accurate identification of vehicles and estimating density in traffic surveillance systems is a challenging task, particularly in scenarios with closely spaced lanes. Single Shot MultiBox Detector (SSD) is introduced in vehicle detection and classification due to its speed and accuracy. It utilizes a transfer learning technique that enables them to utilize features from pretrained Convolutional Neural Networks (CNNs). Although the utilization of multi-scale feature maps in SSD has achieved efficient results in vehicle detection, it struggles to identify key features of vehicles due to its one-stage detection approach. Furthermore, the use of VGG16 as the backbone network in these approaches leads to the loss of fine-grained details, posing challenges in accurately localizing and classifying small vehicles. Also, there is no interaction between high- and low-level features, which restricts the networks ability to effectively integrate and utilize both features for accurate vehicle detection. To overcome these limitations, an improved SSD approach is introduced in this paper, leveraging interactive multi-scale attention characteristics to accurately detect and classify vehicles. This approach utilizes ResNet50 as the network backbone to overcome the limitations of traditional SSDs in detecting small vehicles. Also, an attention block is incorporated into it to focus on key details and assign higher attention to relevant pixels within the feature map. Also, the network employs a parallel detection framework and shares multi-scale layers (both high and low), enabling efficient detection of vehicles with various sizes. Then, traffic density is estimated based on the weights assigned to the categorized vehicles obtained from the improved SSD and the recorded area. Finally, traffic is classified as high, low, or moderate by comparing the estimated density to a threshold value. By adding attention characteristics of different scales to the original detection branch and replacing the VGG16 with ResNet50 of the SSD technique using our method, the feature representation capability and detection accuracy are both significantly improved. [ABSTRACT FROM AUTHOR]

It is common for farmers to detect infections by speaking with regional specialists and conducting further experiments. Farming illnesses that eventually kill crops can impede crop growth. Due to the widespread spread of new diseases, farmers can have trouble understanding what's happening when things stabilize. In addition to affecting the nation's economy, agricultural diseases are crucial to controlling. Agricultural losses in India are 22.13% annually. Machine learning and the cloud can be used to identify diseases automatically to halt further crop losses. Even though a machine learning model is available for this proposal, it is not particularly advanced, reliable, portable, or practical in many ways. The user interface of the website allows farmers to upload a photo of a leaf to a cloud-hosted website to see how the disease has affected their crops. With the help of real-time data generated from 7,000 leaf photos, Convolutional Neural Networks (CNN), MobileNet, and machine learning libraries such as TensorFlow and OpenCV, it is possible to identify plant diseases with an average accuracy of 98.12%. [ABSTRACT FROM AUTHOR]

Textile electrodes are a new and potential choice for long term and continuous monitoring of electrocardiogram measurements. In this research, textile electrodes have been designed and developed by embroidering conductive yarn on the polyester fabric to measure electrocardiogram biosignals of patients. Silver-coated nylon thread is embroidered on 100% polyester fabric in square form 11 mm × 11 mm with satin stitches in order to form three heart vectors (V3, V4, and V5). These electrodes are characterized based on electrode impedance and electrocardiogram measurement and the results are compared with the commercially available disposable Ag/AgCl electrodes. Further, additional tests have been performed using an electrode designed with 50% of the conductive thread used in the original sample. Real-time electrocardiogram signals, QRS-complex, P-wave, and T-wave were obtained for 10 subjects using existing hospital electrocardiogram equipment. The performance and repeatability of the electrode designed with 50% less conductive thread and effect of laundering were also studied. The results showed that the impedance of electrode has an acceptable value of 1.45 MΩ/cm2. Ageing tests showed that there is only a negligible deviation in the performance of the electrode. The results after laundering showed that the performance of the electrode is not affected by the laundering process. [ABSTRACT FROM AUTHOR]

This paper deals with the development of electro-active elastomeric fabric sensor using silver coated polyamide yarn at the centre of the tape along with polyester and rubber yarns and its electrical characterisation for the application towards elbow angle measurement. Box and Behnken 3-variable and 3-level experimental design has been used and 15 different samples are produced. The resistance change in the samples caused by linear extension as well as the angular movement of the elbow is measured. The variables for sensor design have been optimized using gauge factor (Δr/Δ.x) values as dependent variable and the results show that the variables such as number of conductive threads, picks per inch and rubber: polyester thread ratio influence the Δr/Δx value negatively. Different sensor characteristics have also been analysed. The sample with 4 conductive threads, 20 picks/inch and 1:8 polyester : rubber ratio shows a higher gauge factor of 0.583 and has proportional resistance change with length; this sample is therefore chosen to be good among all the 15 samples developed. It is also observed that this sample with higher sensitivity value can be used to measure elbow joint up to 90° angle. [ABSTRACT FROM AUTHOR]