[{"data":1,"prerenderedAt":247},["ShallowReactive",2],{"project-16511940":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":43,"post_text":44,"what":45,"said":69,"maker_in_thread":149,"shape":151,"grid":155,"attempts":208,"same_week":209,"maker_shows":33,"maker_hidden":239,"reader_version":240,"card":241},"16511940","Machine Box Classificationbox","Quickly build a production ready image classifier from your images","title","Show HN: Quickly build a production ready image classifier from your images","https:\u002F\u002Fblog.machinebox.io\u002Fhow-anyone-can-build-a-machine-learning-image-classifier-from-photos-on-your-hard-drive-very-5c20c6f2764f","blog.machinebox.io","matryer",1520114934,2018,142,20,293,2155,14,"notable","ai","image classification","api",[24,25,26,27,28],"image classifier","machine learning","docker","image recognition","training data","curious",-0.091,{"positive":32,"neutral":33,"mixed":34,"critical":35},1,5,0,3,9,{"execution":38,"performance":39,"pricing":41,"idea":42},{"positive":32,"critical":34},{"positive":34,"critical":40},2,{"positive":34,"critical":34},{"positive":34,"critical":32},[34,35,34,34,34,34,34,32,34,34,34,34,35,32,34,32,33,35,35,35],null,{"problem":46,"audience":44,"pricing":44,"motivation":44,"time_to_build":44,"stack":52,"openness":68,"stage":68,"team_size":68},{"value":47,"source":48,"evidence":49},"Training an image classifier on your own photos without building ML from scratch","inferred",[50],{"id":51},"post",[53,61],{"value":54,"source":55,"evidence":56},"Docker","commenter",[57],{"id":58,"by":59,"text":60},"16512558","amelius","Is that Docker image capable of utilizing a GPU?",{"value":62,"source":55,"evidence":63},"Base64 image encoding for API",[64],{"id":65,"by":66,"text":67},"16512150","aisofteng",">Teaching essentially involves opening each image, converting it into a base64 string, and submitting it to the \u002Fclassificationbox\u002Fteach API endpoint in a request that resembles…","unknown",{"praise":70,"objections":86,"requests":115,"questions":128,"alternatives":140},[71],{"value":72,"source":55,"evidence":73},"Looks great and easy to try",[74,78,82],{"id":75,"by":76,"text":77},"16512692","pitchups","This looks great! But not entirely clear when looking at the site, what would be the pricing to classify say 20,000 images from 20 different folders?",{"id":79,"by":80,"text":81},"16512992","smortaz","Looks great! Uploaded to https:\u002F\u002Fnotebooks.azure.com\u002Fsmortaz\u002Flibraries\u002Fhvasslab-TF-tutorials ready to clone & run...",{"id":83,"by":84,"text":85},"16513392","taeric","This is neat, but sadly useless for what I'd imagine many of us have. Pictures of people we want to classify. Possibly of events. Classifying events is often most easily done by…",[87,97,108],{"value":88,"source":55,"evidence":89},"Accuracy not production-grade; classifiers confuse similar images",[90,94],{"id":91,"by":92,"text":93},"16512566","dmitrygr","It is interesting that the words \"production ready\" can be used in connection to image classifiers. Google is unarguably the leader of this in the world, and yet Google photos…",{"id":95,"by":84,"text":96},"16514931","As I said, there are some pictures where the only thing in them are my kids. Lighting was such that they are literally all that is in the picture. And I'm talking about 3 month…",{"value":98,"source":55,"evidence":99},"Base64 encoding adds about a third to image size",[100,104],{"id":101,"by":102,"text":103},"16513949","tonyhb","No, base64 is normally ~30% larger than binary to represent the same data.",{"id":105,"by":106,"text":107},"16513976","JimDabell","Base64 doesn't compress at all, in fact it's only 75% as efficient as binary, meaning that encoding a file into Base64 increases its size by a third. Base64 uses a 64 character…",{"value":109,"source":55,"evidence":110},"Unclear whether data leaves the user's system during training",[111],{"id":112,"by":113,"text":114},"16515397","natch","I think it would be worthwhile for HN to come up with some sort of rule or strong guideline at least that says if you are going to post a Show HN item, you should pop in and…",[116,120,124],{"value":117,"source":55,"evidence":118},"Show pricing for large image sets",[119],{"id":75,"by":76,"text":77},{"value":121,"source":55,"evidence":122},"Tutorial covering custom categories not in standard models",[123],{"id":112,"by":113,"text":114},{"value":125,"source":55,"evidence":126},"GPU support in the Docker image",[127],{"id":58,"by":59,"text":60},[129,136],{"value":130,"source":55,"evidence":131},"How many images are needed to train well",[132],{"id":133,"by":134,"text":135},"16514458","statictype","Do I need a lot of images for this to work? One of the biggest barriers to entry in machine learning is not having enough sample data to train on. I don’t have 10,000 images of…",{"value":137,"source":55,"evidence":138},"Why images are sent as base64",[139],{"id":65,"by":66,"text":67},[141,145],{"value":142,"source":55,"evidence":143},"Google Photos image search",[144],{"id":91,"by":92,"text":93},{"value":146,"source":55,"evidence":147},"Google and Amazon image recognition services",[148],{"id":95,"by":84,"text":96},{"comments":34,"answered":34,"key":150},[],{"commenters":152,"max_depth":33,"maker_replies":34,"first_comment_mins":153,"lifespan_hours":154},18,38,19,[156,158,160,164,167,168,169,172,174,178,180,184,187,189,192,196,197,199,204,206],[112,34,34,157,113,34],"other\u002Fneutral\u002Fdocs",[65,32,35,159,66,34],"project\u002Fneutral\u002Fexecution",[161,32,34,162,163,34],"16512177","commenter\u002Fneutral\u002Fnone","spooneybarger",[165,32,34,162,166,34],"16513914","samhunta",[101,32,34,162,102,34],[105,32,34,162,106,34],[170,32,34,162,171,34],"16514726","gregmac",[91,40,32,173,92,34],"project\u002Fcritical\u002Fperformance",[175,40,34,176,177,34],"16512681","commenter\u002Fneutral\u002Fperformance","omalleyt",[179,40,34,162,92,34],"16512753",[181,40,34,182,183,34],"16514233","commenter\u002Fcritical\u002Fperformance","akvadrako",[185,40,34,176,186,34],"16512823","tensor",[75,35,35,188,76,34],"project\u002Fneutral\u002Fpricing",[83,190,32,191,84,34],4,"project\u002Fcritical\u002Fidea",[193,190,34,194,195,34],"16514242","commenter\u002Fneutral\u002Fidea","Iv",[95,190,32,173,84,34],[79,33,33,198,80,34],"project\u002Fpositive\u002Fexecution",[200,201,35,202,203,34],"16512886",6,"project\u002Fneutral\u002Fidea","jvm_",[133,205,35,202,134,34],7,[58,207,35,159,59,34],8,[],[210,214,219,224,229,234],{"id":211,"name":212,"points":213,"category":20},"16535358","Mathpix",456,{"id":215,"name":216,"points":217,"category":218},"16490595","Graphite",330,"productivity",{"id":220,"name":221,"points":222,"category":223},"16521385","Hearth",315,"infrastructure",{"id":225,"name":226,"points":227,"category":228},"16529557","Investment Calculator",308,"finance_crypto",{"id":230,"name":231,"points":232,"category":233},"16513916","Emacs Anywhere",248,"dev_tools",{"id":235,"name":236,"points":237,"category":238},"16520955","Hacking a Mercedes GLA Class Ride-on kids car",221,"hardware",false,"v2",{"id":242,"name":5,"line":6,"ai":239,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":243,"top":18,"kw":244,"maker":11,"domain":10,"tier":19,"attempts":34,"strip":245,"url":9,"mix":246,"praise":72,"objection":88,"sub":21},16511940,-0.09,[24,25,26,27,28],"03000001000031015333",[32,33,34,35],1793990971261]