Sahil Dadia

SAHIL​ ​MANOJ​ ​DADIA Mobile:​​ ​+353​ ​-​ ​|​ ​E-Mail​:​-Address​:​ ​53​ ​Thornhill​ ​Meadows,​ ​Celbridge,​ ​Ireland Date​ ​of​ ​Birth​:​ ​26​th​​ ​Jul​ ​1994​ ​|​ ​Place​ ​of​ ​Birth​:​ ​Mumbai​ ​|​ ​Citizenship​:​ ​Indian ​ ​COMPUTER​ ​VISION​ ​RESEARCH​ ​EXPERIENCE Position:​ ​Computer​ ​Vision​ ​Research ​ ​ ​Sep​ ​2016​ ​-​ ​Jun​ ​2017 Swaayatt​ ​Robots​,​ ​Autonomous​ ​Car​ ​Technology​ ​Research​ ​|​ ​Advisor​ ​:​ ​Sanjeev​ ​Sharma Swaayatt Robots is focused on building self-driving vehicle technology. This requires investigating entire computer vision literature to identify key algorithms and frameworks for solving outdoor perception problems. Object​ ​Detection​ ​and​ ​Classifications o Implemented​ ​and​ ​tested​ ​8​ ​state​ ​of​ ​the​ ​art​ ​architectures​ ​for​ ​object​ ​detection​ ​and​ ​classification. o Research revealed various flaws of algorithms & as a result new architectures were created to counter​ ​extreme​ ​blurring,​ ​scale​ ​change,​ ​and​ ​obtain​ ​real-time​ ​performance. Object​ ​Tracking o Implemented and benchmarked ​object tracking frameworks such as TLD, KCF, GoTURN etc for highways​ ​and​ ​city​ ​traffic. o New technique was developed to perform occlusion reasoning and make the algorithms computationally​ ​feasible​ ​for​ ​real-time​ ​performance. Automatic​ ​Dataset​ ​Annotation​ ​for​ ​Videos o Created algorithm to mark ​object class agnostic segmentation mask of object, by accounting for pose change without human intervention or external information in Videos. Primarily created to alleviate​ ​need​ ​for​ ​manual​ ​annotations​ ​by​ ​humans. o Current​ ​framework​ ​is​ ​used​ ​for​ ​creation​ ​of​ ​annotated​ ​datasets​ ​for​ ​internal​ ​use​ ​at​ ​Swaayatt​ ​Robots. Converted​ ​4-Wheel​ ​Vehicle​ ​to​ ​Drive​ ​By​ ​Wire o Contributed to design and development of automatic brake, clutch and steering mechanism. It was done​ ​to​​ ​convert​ ​a​ ​manual-transmission​ ​vehicle​ ​to​ ​an​ ​autonomous​ ​vehicular​ ​platform​. Multi​ ​Camera​ ​-​ ​Multi​ ​View​ ​Camera​ ​Calibration o Wrote programs for robust camera calibration of 4 camera system (internal-external) located at various​ ​locations​ ​on​ ​a​ ​vehicle. Position​ ​:​ ​Computer​ ​Vision​ ​Research ​ ​Dec​ ​2015​ ​–​ ​Feb​ ​2016 Swaayatt​ ​Robots​,​ ​Autonomous​ ​Car​ ​Technology​ ​Research​ ​|​ ​Advisor:​ ​Sanjeev​ ​Sharma Swaayatt Robots is focused on building self-driving vehicle technology. This requires investigating entire computer vision literature to identify key algorithms and frameworks for solving outdoor perception problems. Lane​ ​Marker​ ​and​ ​Object​ ​Saliency​ ​Detection​ ​in​ ​Images o Investigated and implemented latest research for ​lane marker detection providing insight for various failure cases and their causes for severe Indian Road Conditions. Moreover, it became the base​ ​for​ ​further​ ​lane​ ​marker​ ​and​ ​lane​ ​departure​ ​safety​ ​systems​ ​developed​ ​at​ ​Swaayatt​ ​robots. o Object saliency detection was implemented to assist Object Detection and Classification. Assessment of causes of failure of saliency detection methods prompted for further research directions​ ​at​ ​Swaayatt​ ​Robots. EDUCATION MSc​ ​in​ ​Data​ ​Science​ ​and​ ​Analytics​ ​(Pursuing) Maynooth​ ​University,​ ​Ireland ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​Sep​ ​2017​ ​-​ ​Present Bachelor​ ​of​ ​Engineering​ ​(Mechanical​ ​Branch)​ ​-​ ​CGPA:​ ​7.8/10.0 Dwarkadas​ ​J.​ ​Sanghvi​ ​College​ ​of​ ​Engineering,​ ​University​ ​of​ ​Mumbai,​ ​India ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​Jul​ ​2012​ ​–​ ​Jul​ ​2016 ​ ​PROJECTS​ ​(self​ ​initiated) Insight​ ​for​ ​BI ​ ​ ​ ​ ​Oct​ ​2017​ ​-​ ​Present Deep​ ​Learning​ ​&​ ​Optimizations​ ​Library ​ ​ ​ ​ ​Jun​ ​2017​ ​-​ ​Present Vision​ ​Based​ ​Semi-Autonomous​ ​Quadcopter Jun​ ​2015​ ​–​ ​Mar​ ​2016 Self​ ​Initiated o SaaS for small and medium enterprise regarding the customer pattern recognition and targeting to maximise​ ​sales​ ​and​ ​minimize​ ​inventory​ ​costs o Removes​ ​the​ ​pain​ ​points​ ​in​ ​using​ ​the​ ​current​ ​BI​ ​softwares​ ​currently​ ​present​ ​in​ ​the​ ​market Self​ ​Initiated o Aim to create a framework for optimization/deep learning research using C++ without explicit need to​ ​write​ ​a​ ​backward​ ​propagation​ ​code. Final​ ​year​ ​(B.E)​ ​project​ ​|​ ​Advisors:​ ​Prashantkumar​ ​Patankar​ ​and​ ​Pavankumar​ ​Borra o Aim was to detect objects from top-view in quadcopters and follow the fastest object autonomously o Challenges included separation of ​ego-motion, trajectory planning and stable tracking​. Additionally, stress analysis and motion simulation was performed to assess the stability and manoeuvrability of the​ ​quadcopter​ ​frame ​ ​TECHNICAL​ ​PROFICIENCY TECHNOLOGY EXPERIENCE Programming​ ​Languages ● Python ● C/C++ 3​ ​years Image​ ​Processing​ ​/​ ​Computer​ ​Vision ● OpenCV ● DLib ● Caffe ● MXNet 2​ ​years Machine​ ​Learning​ ​/​ ​Deep​ ​Learning ● Caffe ● MXNet ● Theano ● Keras ● Scikit-learn ● Scipy ● Numpy 2​ ​years ​ ​MATHEMATICAL​ ​COURSES Course​ ​Topic​ ​-​ ​PostGrad​ ​Level Machine​ ​Learning Pattern​ ​Recognition Convolutional​ ​Neural​ ​Networks Multi​ ​View​ ​Geometry Professor Andrew​ ​Ng Sukhendu​ ​Das Andrej​ ​Karpathy Daniel​ ​Cremers Organization Stanford​ ​University IIT​ ​Madras​ ​(NPTEL) Stanford​ ​University Technical​ ​University​ ​Munich Course​ ​Topic​ ​-​ ​PostGrad​ ​Level Machine​ ​Learning Pattern​ ​Recognition Convolutional​ ​Neural​ ​Networks Multi​ ​View​ ​Geometry Professor Andrew​ ​Ng Sukhendu​ ​Das Andrej​ ​Karpathy Daniel​ ​Cremers Organization Stanford​ ​University IIT​ ​Madras​ ​(NPTEL) Stanford​ ​University Technical​ ​University​ ​Munich ​ ​ ​ ​PERSONAL​ ​DOSSIER Languages​ ​:​ ​ ​English,​ ​Hindi,​ ​Gujrati,​ ​Marathi,​ ​German​ ​(A1​ ​German​ ​Exam) Date​ ​of​ ​Birth​ ​:​ ​26​ ​July​ ​1994