[{"data":1,"prerenderedAt":302},["ShallowReactive",2],{"project-28100346":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":29,"sentiment":30,"stance_mix":31,"scored_comments":36,"aspects":37,"strip":45,"post_text":47,"what":48,"said":89,"maker_in_thread":170,"shape":181,"grid":183,"attempts":265,"same_week":266,"maker_shows":32,"maker_hidden":294,"reader_version":295,"card":296},"28100346","Stray Robots toolkit","Object Detection in an Hour","title","Show HN: Object Detection in an Hour","https:\u002F\u002Fwww.strayrobots.io\u002Fblog\u002Fobject-detection-in-an-hour","strayrobots.io","kekeblom",1628357889,2021,166,34,289,2533,12,"notable","ai","computer vision toolkit","library",[24,25,26,27,28],"object detection","computer vision","depth camera","3d annotation","training data","curious",-0.011,{"positive":32,"neutral":33,"mixed":34,"critical":35},2,14,1,3,20,{"idea":38,"performance":39,"design":41,"execution":42,"docs":43,"pricing":44},{"positive":34,"critical":34},{"positive":40,"critical":40},0,{"positive":40,"critical":40},{"positive":40,"critical":32},{"positive":40,"critical":40},{"positive":40,"critical":40},[40,46,40,35,35,35,35,40,35,35,40,40,34,40,40,40,34,40,40,35,35,35,34,46,40,35,35,35,35,40,40,35,40,35],5,null,{"problem":49,"audience":47,"pricing":56,"motivation":62,"time_to_build":47,"stack":66,"openness":86,"stage":87,"team_size":88},{"value":50,"source":51,"evidence":52},"Computer vision projects take months; training data and tooling are hard to get working","maker",[53],{"id":54,"by":11,"text":55},"28100353","Hi HN! hietalajulius and I have been working on a toolkit for solving computer vision problems. These days, there are a lot of fancy solutions to many computer vision problems,…",{"value":57,"source":51,"evidence":58},"Plans to charge for some algorithms and cloud compute credits",[59],{"id":60,"by":11,"text":61},"28101606","We plan to charge for some of the algorithms. Also, eventually some parts will run in our cloud and we could charge for compute credits.",{"value":63,"source":51,"evidence":64},"Lower the barrier to deploying computer vision solutions",[65],{"id":54,"by":11,"text":55},[67,75,82],{"value":68,"source":69,"evidence":70},"Detectron2","commenter",[71],{"id":72,"by":73,"text":74},"28107248","hietalajulius","Yeah the labels are loaded into the Detectron2 format from the 3D annotations json at train time, we plan to add similar data loading for YOLOv3 etc soon. Starting out with…",{"value":76,"source":69,"evidence":77},"PyTorch",[78],{"id":79,"by":80,"text":81},"28101492","AndrewKemendo","Just so I understand the idealized pipeline here, a user does the following: 1. Use the Scanner app to take the images and camera pose data 2. Export the scene directory (color…",{"value":83,"source":69,"evidence":84},"TorchScript model export",[85],{"id":79,"by":80,"text":81},"unknown","beta","team",{"praise":90,"objections":104,"requests":123,"questions":135,"alternatives":154},[91,98],{"value":92,"source":69,"evidence":93},"Clever use of video to build large training sets",[94],{"id":95,"by":96,"text":97},"28101236","jonatron","Using video to automatically build a large training set is smart! Well done! I was thinking about making a properly free and open dataset from just walking around London, and…",{"value":99,"source":51,"evidence":100},"Impressive detection of moving objects without training on them",[101],{"id":102,"by":11,"text":103},"28102345","Yes it relies on the target being static when capturing the training data, but it’s ok for the background to move. We were actually surprised by how well it works on moving…",[105,112],{"value":106,"source":69,"evidence":107},"Only works when the target object is static during capture",[108],{"id":109,"by":110,"text":111},"28102175","tadeegan","This is only beneficial in static scenes right? Otherwise you can’t get free labels across the whole video.",{"value":113,"source":69,"evidence":114},"iPhone app requires LiDAR-equipped Pro models; other iPhones are unsupported",[115,119],{"id":116,"by":117,"text":118},"28101330","posix_compliant","Heads up for anyone else, I was interested in the strayscanner app to try on my iPhone 11, but I’m getting an error when trying to record: “unsupported device: this device…",{"id":120,"by":121,"text":122},"28103155","TomVDB","I just got caught by this myself: it doesn't work on an regular iPhone 12 either. It needs to be an iPhone 12 Pro or 12 Pro Max.",[124,131],{"value":125,"source":69,"evidence":126},"Support front TrueDepth camera",[127],{"id":128,"by":129,"text":130},"28103216","pzo","Maybe worth to support front camera (TrueDepth camera) as well? Record3d gives pretty good accuracy (https:\u002F\u002Frecord3d.app\u002F). I know probably not the best way to scan something…",{"value":132,"source":69,"evidence":133},"Support YOLO and other model formats beyond Detectron2",[134],{"id":72,"by":73,"text":74},[136,147],{"value":137,"source":69,"evidence":138},"Is 3D labelling actually faster than 2D annotation tools",[139,143],{"id":140,"by":141,"text":142},"28101593","rocauc","Do you comprehend how the tool reduces labeling time in (4) and (5) (compared to labeling with eg CVAT) as the post claims?",{"id":144,"by":145,"text":146},"28103650","mainframed","You only have to label the 3D Bounding Box once. Then you can automatically generate 2D bounding boxes for every frame of the video. So instead of annotating every frame with a…",{"value":148,"source":69,"evidence":149},"What is the business plan",[150],{"id":151,"by":152,"text":153},"28101529","nathan_phoenix","Just curious, what's the business plan behind this?",[155,159,163],{"value":156,"source":69,"evidence":157},"CVAT annotation tool",[158],{"id":140,"by":141,"text":142},{"value":160,"source":69,"evidence":161},"Record3D",[162],{"id":128,"by":129,"text":130},{"value":164,"source":69,"evidence":165},"SiteScape LiDAR 3D scanner",[166],{"id":167,"by":168,"text":169},"28103026","bredren","I was surprised to read this, but it appears you are correct. This other app uses the entire first paragraph to warn about this. Maybe something to copy:…",{"comments":171,"answered":172,"key":173},8,4,[174,175,178],{"id":54,"text":55},{"id":176,"text":177},"28101544","Pretty much yeah. Just to be clear, we only use the color and depth images from the camera. There is actually an offline calibration step to obtain camera intrinsic parameters,…",{"id":179,"text":180},"28104725","Yes if you only care about 2d bbox detection, a smart bounding box annotation tool has some advantages. If you need to solve 3D vision tasks, as is the case in 3D bbox detection,…",{"commenters":36,"max_depth":46,"maker_replies":171,"first_comment_mins":34,"lifespan_hours":182},21,[184,186,187,191,194,198,200,202,203,205,207,208,210,212,214,218,221,224,227,230,232,234,235,236,241,243,246,249,250,252,256,258,261,263],[54,40,40,185,11,34],"project\u002Fpositive\u002Fidea",[95,34,46,185,96,40],[188,34,40,189,190,40],"28101371","commenter\u002Fneutral\u002Fnone","drooby",[192,34,35,193,73,40],"28106998","project\u002Fneutral\u002Fnone",[195,32,35,196,197,40],"28102831","project\u002Fneutral\u002Fperformance","actually_a_dog",[199,32,35,196,73,40],"28107181",[79,35,35,201,80,40],"project\u002Fneutral\u002Fdesign",[176,35,40,201,11,34],[140,35,35,204,141,40],"project\u002Fneutral\u002Fidea",[144,35,35,206,145,40],"project\u002Fmixed\u002Fidea",[179,35,40,204,11,34],[209,35,40,189,145,40],"28105738",[109,172,34,211,110,40],"project\u002Fcritical\u002Fidea",[102,172,40,213,11,34],"project\u002Fpositive\u002Fperformance",[215,46,40,216,217,40],"28103241","other\u002Fneutral\u002Fnone","kayhi",[219,46,40,216,220,40],"28103411","mlboss",[116,222,34,223,117,40],6,"project\u002Fcritical\u002Fexecution",[225,222,40,226,11,34],"28101358","maker\u002Fneutral\u002Fexecution",[228,222,40,229,11,34],"28101378","maker\u002Fneutral\u002Fnone",[231,222,35,226,117,40],"28101493",[167,222,35,233,168,40],"maker\u002Fneutral\u002Fdocs",[128,222,35,201,129,40],[120,222,34,223,121,40],[237,238,46,239,240,40],"28102880",7,"maker\u002Fpositive\u002Fnone","luisramalho",[242,238,40,189,73,40],"28107082",[244,171,35,201,245,40],"28101655","notum",[247,171,35,201,248,40],"28103776","wokwokwok",[72,171,35,201,73,40],[251,171,35,201,73,40],"28107128",[253,254,40,216,255,40],"28101233",9,"tvirosi",[257,254,40,229,11,34],"28101255",[151,259,35,260,152,40],10,"maker\u002Fneutral\u002Fpricing",[60,259,40,262,11,34],"project\u002Fneutral\u002Fpricing",[264,259,35,260,152,40],"28105338",[],[267,272,277,281,285,289],{"id":268,"name":269,"points":270,"category":271},"28074827","Repo Visualization",283,"dev_tools",{"id":273,"name":274,"points":275,"category":276},"28088882","GrowthBook",199,"data",{"id":278,"name":279,"points":280,"category":20},"28120247","colab.research.google.com",169,{"id":282,"name":283,"points":284,"category":271},"28084104","Systemizer",167,{"id":286,"name":287,"points":288,"category":271},"28128445","Lona",165,{"id":290,"name":291,"points":292,"category":293},"28085920","Paper Time",161,"learning_science",false,"v2",{"id":297,"name":5,"line":6,"ai":294,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":298,"top":18,"kw":299,"maker":11,"domain":10,"tier":19,"attempts":40,"strip":300,"url":9,"mix":301,"praise":92,"objection":106,"sub":21},28100346,-0.01,[24,25,26,27,28],"0503333033001000100333150333300303",[32,33,34,35],1793991031556]