Darwin Steni

Experience

• 

  IBM India Private Limited

  Dec 2013 - Oct 2016

  Application Architect

  Language : Python, Shell ScriptingSAP : SAP R3, S/4 HANADatabase : Oracle 10g, Oracle 11g, Oracle 12c (with ASM), Operating Systems : Linux, Unix, Windows XP/7/8/10

• 

  Bosch

  Oct 2016 - now

  •Investigated and Implemented an Efficient Reverse Image Search of over 1M images from the BOSCH Image Library with calculated accuracy of 81% (mAP).•Applied Transfer Learning from Pre-Trained Models to adapt for the images better to fit into the defined problem statements. •Built an efficient and fast Image retrieval system using Annoy tree and HNSW graph indexing methods.•Developed a Web UI to facilitate feedbacks on resultant images using Flask and Python for evaluating the implemented models.•Investigated and Planned the approach for typical VMMR for 3D bounding box image tagging for ADAS project.•Implemented Scene Text Localization approach using pre-trained EAST and CRAFT models.•Implemented and Trained a Custom Dataset for multiclass Object Detection using Single Shot Detection algorithm with ResNet as backbone for BOSCH ADAS project.•Developed a Computer Vision Approach for BSH products assembly with AR Solution.•Currently researching an efficient and generalized edge detection and matching approach using multi-level image smoothening. Show less Fundamentals: Image formation; acquisition, perception; representation; pixel relationshipsImage compression: subband coding and wavelets; fractals;Image enhancement and restoration: image transforms; order-statistics filters; smoothing and sharpening; projection on convex sets; reconstruction from Incomplete dataImage segmentation: image features; texture analysis; clustering algorithmsImage compression: Discrete Cosine Transform; Discrete Fourier transform; Lossy and lossless image compressionMorphology: image blurring; sharpening; edge detection; shape analysisImage pattern recognition: image fusion; image security; image steganography; image compositing3D motion analysis: rigid objects; articulate objects; non-rigid motionOptical image processing: 2D Fourier transform; linear filtering; nonlinear processingFrom image/video databases: content-based and similarity-based indexing and retrieval; video segmentation and keyframe extraction Show less

  • Computer Vision Research Engineer

    Oct 2016 - now

  • Technical Specialist

    Oct 2016 - now