Our Prize Winning AWS Hackathon Entry

Aus Vokipedia
Version vom 17. August 2025, 23:20 Uhr von LeonelBarney698 (Diskussion | Beiträge)

(Unterschied) ← Nächstältere Version | Aktuelle Version (Unterschied) | Nächstjüngere Version → (Unterschied)
Wechseln zu: Navigation, Suche


This article is written to explain my undertaking that I submitted to the AWS Hackathon on Devpost. Me and my friend had preliminary began the challenge as a learning step to build and deploy a machine learning venture. We ultimately ended up winning the runner up for Computer Vision for our efforts. With speedy urbanization and industrialization Air pollution has become an alarming concern globally. Among different air pollutants particles , matter in (PM2.5) size vary are able traverse deeply into our respiratory tract , BloodVitals SPO2 transmitting hazardous chemicals into the human lungs and blood causing respiratory and cardiovascular health points . We purpose to develop picture-primarily based air high quality analysis application , in particular a one that estimates the concentration of particulate matter with diameters in vary of 2.5 micrometers. It may let you know what your respiratory from images generated using shopper grade hardware. Convolutional Neural Network (CNN) with the excessive computational energy of Amazon EC2 DL1 instance to resolve a computer vision drawback of classifying pure photos into 3 different categories ;wholesome , average and hazard based on their PM2.5 concentrations.



We created our personal dataset by combining knowledge from 2 sources . We discovered that, some causes for overfitting could possibly be high imbalance in data , we had 2232 wholesome pictures where because the average and hazard categories had solely 1480 and 591 pictures respectively . 2232. Then we skilled our mannequin following the identical method as before but this time we used ImageNet weights un-freezed the final one hundred layers for coaching , then high-quality-tuned the mannequin with a lower studying price ,we additionally used early stopping callback with a persistence of 3 . Now, there are some more issues we've performed to enhance our mannequin. We migrated our mannequin onto the dl1 occasion and initially educated utilizing 1 gaudi processor. This considerably improved our performance and effectivity .We have been capable of rapidly check , experiment and tweak adjustments . 76 %. We then converted our model into a tensorflow lite mannequin and used a huggingface demo device known as gradio to shortly exhibit our application.



Initially as we tried coaching our mannequin we noticed a few of the picture information in the dataset have been corrupted .We automated the strategy of scanning by means of and deleting these files and BloodVitals SPO2 then proceeded to coaching mannequin . We used EfficientNet structure since its renown for its efficiency in low spec machines, this known as transfer learning . 30 layers and BloodVitals SPO2 set the weights parameter to false. 60 % however our coaching accuracy was high, the model could be overfitting. As we have now seen we had been ready an accuracy of about 76 % which is fairly good contemplating the info we had. Now , why do we need this utility and how is it different from the existing air monitoring system methods, we already know Exposure to tremendous particles may cause long term health effects akin to lung and heart drawback however they also cause short time period well being effects comparable to eye, nose, throat irritation.



The present air quality monitoring strategies are highly dependent on monitoring stations, BloodVitals SPO2 that are again situated far away due to the high setup cost and expensive equipment. For a selected location, the air quality depends upon how air moves by means of the realm and the way individuals are influencing the air , so we cannot rely on something that was predicted a while again or in some place that isn't our site of interest .This application helps you analyse the air high quality around you with a single picture and more importantly is portable. We will carry it anyplace we want to and know what your being uncovered to . Honestly I don’t know, I feel there may be potential for this app to integrate into our routine. People may need options to assist plan their day out, integrate with their phone’s calender and counsel the very best time to plan an out of doors exercise. Right now the Artificial intelligence that runs the machine runs on the azure cloud. In the future we wish to be able to bring it into the app and run it natively.



Certain constituents within the blood have an effect on the absorption of light at numerous wavelengths by the blood. Oxyhemoglobin absorbs light more strongly within the infrared area than in the purple area, BloodVitals SPO2 whereas hemoglobin exhibits the reverse conduct. Therefore, highly oxygenated blood with a excessive focus of oxyhemoglobin and a low concentration of hemoglobin will tend to have a excessive ratio of optical transmissivity in the crimson region to optical transmissivity within the infrared region. These alternating portions are amplified after which segregated by sampling devices working in synchronism with the pink/infrared switching, so as to provide separate alerts on separate channels representing the purple and infrared mild transmission of the body construction. After low-move filtering to remove sign parts at or above the switching frequency, each of the separate alerts represents a plot of optical transmissivity of the body structure at a particular wavelength versus time. AC part brought on only by optical absorption by the blood and varying on the pulse frequency or coronary heart charge of the organism.

Meine Werkzeuge
Namensräume

Varianten
Aktionen
Navigation
Werkzeuge