This third-year engineering pupil created laptop imaginative and prescient software program to stop harassment by means of

What if one have been to let you know that Huge Brother is usually a ‘good’ brother? However the privateness issues and debates round surveillance cameras, metropolitan cities as we speak are coated with them. But, more often than not, there is no such thing as a one governing motion by means of the digicam to catch crimes occurring in real-time. This stands true of India with its excessive sexual abuse instances, lots of which occur in public locations in broad daylight. Souvik Ghosh, a third-year engineering pupil, has created a mannequin, ‘Sathi’, an automatic defender of harassment in public locations. The prototype reveals a lady exhibiting a fist or the index finger to any surveillance digicam on the road and the digicam alerting bypassers with a loud siren and noticing the police, all by itself. In an interview with Analytics India Journal, Souvik spoke about Sathi and his life as a pupil information scientist.  

https://www.linkedin.com/embed/feed/replace/urn:li:ugcPost:6909446808335257600

AIM: What was the concept behind the Sathi mannequin?

My pursuits lie in information science, laptop imaginative and prescient and deep studying. Together with being a tech fanatic, I’m additionally a social employee. I work within the Kolkata crimson gentle space and train the scholars there. This has allowed me to see the abuse that occurs first handed. Later, I discovered that one in each three girls has skilled sexual harassment in India. My feminine buddies additionally spoke about such experiences, and I realised harassment normally occurs in locations like busy roads, crowded buses, native trains, instructional establishments and extra. However many ladies really feel hesitant to convey it to note or shout in public locations. They fail to react. Trying to find options for this made me realise that metropolitan cities like Mumbai, Delhi or Kolkata have surveillance cameras across the metropolis – even transport, streets and faculties. Moreover, most individuals carry smartphones on them all the time. However nobody is actually checking the digicam footage to determine such crimes. I realised we would have liked the digicam to be automated, and it led me to create Sathi.

AIM: Are you able to run us by means of Sathi?

Let’s suppose anybody runs right into a scenario the place they really feel threatened. They should present their fist in an indication of protest in the direction of the digicam, be it the CCTV digicam and even their smartphone digicam. The machine will begin taking part in a loud alarm or siren sound upon registering the fist. This might alert the residents round. Alternatively, if the particular person reveals their index finger or factors to primary, the system will straight inform the police. It should additionally seize and ship the photographs of that spot with a time graph to the police. The system will seize each face in that scene as passport-sized faces as proof to make use of later. For the app, the particular person can present a fist to the cell phone digicam, and the telephone will straight name the police. 

AIM: Inform us about your tech stack. Moreover, as a pupil information scientist, how did you entry the wanted sources?

Due to the web, we now have nice free programs on YouTube and Coursera. The schooling sources are straightforward. On the subject of the system, we dont want entry to excellent methods. We have now Kaggle and Google Collab that present large RAMs. Even in case you keep within the remotest elements of the nation with the most cost effective laptops, you possibly can contribute to system creation.

My technological stack was primarily deep studying, laptop imaginative and prescient and information science. My base language was Python. I used TensorFlow and MediaPipe for Face detection for laptop imaginative and prescient, to make sure the system may detect faces even in low gentle areas. 

AIM: What’s the scope of the mannequin? 

The system remains to be a prototype as of now. The plan is to embed it in a CCTV digicam and a cellular digicam. There are a couple of solutions I’m nonetheless engaged on. Firstly, the space between the particular person and the digicam shouldn’t be all the time going to be ten to fifteen meters. Typically it is going to be fifty meters. So, how nicely will the digicam detect faces from a long-distance? Secondly, how nicely will it carry out for a number of faces? For example, how will it determine the face in query from a whole lot of individuals? Lastly, what if folks use it for enjoyable? If I have been a toddler, I’d get pleasure from exhibiting a fist to the digicam and watching the gang create chaos. I plan on connecting with trade specialists and professors to brainstorm and make it foolproof. 

AIM: How do you propose to take the mannequin additional?

My two approaches as of now are utilizing MEMs and embedded computer systems. We will put the entire system in embedded computer systems and CCTV cameras. We will additionally put the code in cellular apps that work in rural areas with out cameras. The app runs within the background and can determine the fist within the digicam. A problem right here is to attach the digicam to gentle posts within the space since a siren in a cell phone is pointless. Upon seeing a fist, the cellular app will notify the sunshine submit that may play a siren. We nonetheless need to brainstorm the place the ultimate model will work; a CCTV digicam, an app or each? We will have city and rural protection by means of each. 

AIM: What are your ideas for college students getting into information science?

Anybody getting into information science ought to begin with Python and its libraries. There are steps one must climb up; folks shouldn’t bounce straight into deep studying. Knowledge science is the basis of all studying, adopted by machine studying and deep studying. That is the trail I adopted. I’ve additionally interned with totally different organisations in positions like a analysis intern. Internships additionally compel me to check and study extra. Lastly, making use of without cost programs on platforms like YouTube will assist research higher.