lightly-studio 0.3.1__py3-none-any.whl → 0.3.2__py3-none-any.whl

lightly_studio/dist_lightly_studio_view_app/_app/immutable/nodes/6.Bw5ck4gK.js

@@ -1 +0,0 @@
-import{r as e}from"../chunks/H7C68rOM.js";import{l as a}from"../chunks/BJXwVxaE.js";import{r as s}from"../chunks/B90CZVMX.js";const l=async()=>{const{data:t,error:o}=await a(),r=t==null?void 0:t.dataset_id;if(o||!r)throw o;s(307,e.toSamples(r))},p=Object.freeze(Object.defineProperty({__proto__:null,load:l},Symbol.toStringTag,{value:"Module"}));export{p as universal};

lightly_studio/dist_lightly_studio_view_app/_app/immutable/nodes/7.CF0EDTR6.js

@@ -1 +0,0 @@
-import"../chunks/CWj6FrbW.js";import{p as f,u as $,a as h,g as o,d as r}from"../chunks/l7KrR96u.js";import{s as S,a as i}from"../chunks/XO7A28GO.js";import{A as T}from"../chunks/CvOmgdoc.js";import{u as _}from"../chunks/vC4nQVEB.js";import"../chunks/CxtLVaYz.js";import"../chunks/D6su9Aln.js";import"../chunks/Bx1xMsFy.js";import"../chunks/Bsi3UGy5.js";import"../chunks/69_IOA4Y.js";import{u as G}from"../chunks/DIRAtgl0.js";import"../chunks/BJXwVxaE.js";function k(p,t){f(t,!0);const[e,m]=S(),n=()=>i(l,"$lastGridType",e),s=()=>i(d,"$sampleSize",e),{datasetId:a,sampleSize:d,selectedAnnotationFilterIds:c}=t.data,{lastGridType:l}=_(),u=r(()=>G({dataset_id:a})),g=r(()=>o(u).clearTagsSelected);$(()=>{n()!=="annotations"&&o(g)()}),T(p,{get itemHeight(){return s().height},get itemWidth(){return s().width},dataset_id:a,selectedAnnotationFilterIds:c}),h(),m()}export{k as component};

lightly_studio/dist_lightly_studio_view_app/_app/immutable/nodes/8.Cw30LEcV.js

@@ -1 +0,0 @@
-import{p as T,f as B,n as Et,a as j,w as bt,l as V,$ as ct,g as a,d,c as f,r as m,s as S,t as F,L as Dt,V as Tt,ax as gt}from"../chunks/l7KrR96u.js";import{c as jt}from"../chunks/BJXwVxaE.js";import"../chunks/CWj6FrbW.js";import{c as Z,a as _,t as k,n as Ht}from"../chunks/Bu7uvVrG.js";import{i as O,a as D,s as q}from"../chunks/XO7A28GO.js";import{e as ut}from"../chunks/DjfY96ND.js";import{I as Rt,s as J,K as Ut,L as at,M as Ft,N as Ot,O as zt,P as Gt,p as xt,F as Kt,e as Wt,l as qt,h as Vt,C as It,z as At,E as wt,u as Zt,d as Jt,S as Qt,Q as Yt,a as Xt}from"../chunks/CvOmgdoc.js";import{g as tt}from"../chunks/CxtLVaYz.js";import{u as St}from"../chunks/vC4nQVEB.js";import{f as ht,S as ta}from"../chunks/D5-A_Ffd.js";import{r as U}from"../chunks/H7C68rOM.js";import"../chunks/D6su9Aln.js";import"../chunks/Bx1xMsFy.js";import{p as z}from"../chunks/Bsi3UGy5.js";import"../chunks/69_IOA4Y.js";import{S as aa}from"../chunks/B2FVR0s0.js";import{S as na,B as ea,a as oa,b as X,c as lt,H as sa,d as dt,D as ra,e as ia,Z as la,g as da,u as ca}from"../chunks/nAHhluT7.js";import{S as ua}from"../chunks/BylOuP6i.js";import{s as va,r as ma,a as fa}from"../chunks/D98V7j6A.js";import{s as W}from"../chunks/D6RI2Zrd.js";import{C as _a}from"../chunks/CDnpyLsT.js";import{s as pa}from"../chunks/DIRAtgl0.js";function ga(v,t){T(t,!0);let r=ma(t,["$$slots","$$events","$$legacy"]);Rt(v,va({name:"square-dashed"},()=>r,{iconNode:[["path",{d:"M5 3a2 2 0 0 0-2 2"}],["path",{d:"M19 3a2 2 0 0 1 2 2"}],["path",{d:"M21 19a2 2 0 0 1-2 2"}],["path",{d:"M5 21a2 2 0 0 1-2-2"}],["path",{d:"M9 3h1"}],["path",{d:"M9 21h1"}],["path",{d:"M14 3h1"}],["path",{d:"M14 21h1"}],["path",{d:"M3 9v1"}],["path",{d:"M21 9v1"}],["path",{d:"M3 14v1"}],["path",{d:"M21 14v1"}]],children:(n,o)=>{var s=Z(),p=B(s);pa(p,()=>t.children??Et),_(n,s)},$$slots:{default:!0}})),j()}const ha=async({dataset_id:v,cursor:t=0,limit:r=10,annotation_label_ids:i,tag_ids:n})=>{const o={data:[],error:void 0};try{const s=await jt.GET("/api/datasets/{dataset_id}/annotations",{params:{path:{dataset_id:v},query:{annotation_label_ids:i,tag_ids:n,cursor:t,limit:r}}});if(s.error)throw new Error(JSON.stringify(s.error,null,2));if(!s.data)throw new Error("No data");o.data=s.data.data}catch(s){o.error="Error loading annotations: "+String(s)}return o},ba=({annotationIndex:v,...t})=>{const r=bt({isLoading:!1});return(async()=>{r.update(n=>({...n,isLoading:!0}));try{const n=await ha({cursor:v>0?v-1:0,limit:3,...t}),o=n.data.findIndex(e=>e.annotation_id===t.currentAnnotationId),s=n.data[o+1],p=n.data[o-1];r.update(e=>({...e,isLoading:!1,error:n.error?String(n.error):void 0,data:n.error?void 0:{annotationNext:s,annotationPrevious:p}}))}catch(n){r.update(o=>({...o,isLoading:!1,error:String(n)}))}})(),r},xa=async({parent:v,params:{dataset_id:t,annotationId:r,annotationIndex:i}})=>{const{annotationsSelectedTagsIds:n,annotationsSelectedAnnotationLabelsIds:o,dataset:s}=await v(),p=parseInt(i,10),e=bt();if(!s)throw new Error("Dataset data not found");return ba({annotationIndex:p,dataset_id:t,currentAnnotationId:r,annotation_label_ids:Array.from(V(o)),tag_ids:Array.from(V(n))}).subscribe(({data:L})=>{e.set(L)}),{annotationAdjacents:e,annotationIndex:p,annotationId:r,dataset:s}},gn=Object.freeze(Object.defineProperty({__proto__:null,load:xa},Symbol.toStringTag,{value:"Module"}));function Ia(v,t){T(t,!0);const[r,i]=q(),n=()=>D(a(s),"$annotationAdjacents",r),o=d(()=>z.data.annotationIndex),s=d(()=>z.data.annotationAdjacents),p=()=>{n().annotationNext&&tt(U.toSampleWithAnnotation({datasetId:n().annotationNext.dataset_id,sampleId:n().annotationNext.sample_id,annotationId:n().annotationNext.annotation_id,annotationIndex:a(o)+1}))},e=()=>{n().annotationPrevious&&tt(U.toSampleWithAnnotation({datasetId:n().annotationPrevious.dataset_id,sampleId:n().annotationPrevious.sample_id,annotationId:n().annotationPrevious.annotation_id,annotationIndex:a(o)-1}))},L=g=>{switch(g.key){case"ArrowRight":p();break;case"ArrowLeft":e();break}};var x=Z();ut("keydown",ct,L);var $=B(x);{var I=g=>{const P=d(()=>!!n().annotationPrevious),y=d(()=>!!n().annotationNext);ua(g,{get hasPrevious(){return a(P)},get hasNext(){return a(y)},onPrevious:e,onNext:p})};O($,g=>{n()&&g(I)})}_(v,x),j(),i()}const Aa={object_detection:"Object Detection",bbox:"Bounding Box",instance_segmentation:"Instance Segmentation",semantic_segmentation:"Semantic Segmentation",classification:"Classification"};var wa=k('<span class="text-sm"> </span> <span class="break-all text-sm"> </span>',1);function Sa(v,t){var r=wa(),i=B(r),n=f(i,!0);m(i);var o=S(i,2),s=f(o,!0);m(o),F(()=>{W(n,t.label),J(o,"data-testid",`annotation-metadata-${t.id}`),W(s,t.value)}),_(v,r)}var $a=k('<span class="break-all text-sm"> </span>'),ya=k('<span class="text-sm"> </span> <!>',1);function ka(v,t){T(t,!0);const[r,i]=q(),n=()=>D(s,"$result",r),o=()=>D(t.isEditingMode,"$isEditingMode",r),s=Ut(),{updateAnnotation:p}=at({datasetId:t.datasetId,annotationId:t.annotationId,onUpdate:t.onUpdate}),e=d(()=>zt(n().data||[])),L=d(()=>{const A=a(e).find(c=>c.value===t.value);return A||{value:t.value,label:t.value}});var x=ya(),$=B(x),I=f($,!0);m($);var g=S($,2);{var P=A=>{Ft(A,{get items(){return a(e)},get value(){return a(L)},name:"annotation-label",placeholder:"Select a label",onSelect:l=>{p({annotation_id:t.annotationId,dataset_id:t.datasetId,label_name:l.value})},notFound:(l,u)=>{let h=()=>u==null?void 0:u().inputValue;Ot(l,{get label(){return h()}})},$$slots:{notFound:!0}})},y=A=>{var c=$a();J(c,"data-testid","annotation-metadata-label");var l=f(c,!0);m(c),F(()=>W(l,t.value)),_(A,c)};O(g,A=>{o()?A(P):A(y,!1)})}F(()=>W(I,t.label)),_(v,x),j(),i()}var Pa=k('<div class="flex flex-col gap-4"><div class="grid grid-cols-[6rem_1fr] gap-y-3 text-diffuse-foreground"></div></div>');function Ma(v,t){T(t,!0);const[r,i]=q(),n=()=>D(a(p),"$annotationResp",r),{datasetId:o}=z.data,s=d(()=>at({datasetId:o,annotationId:t.annotationId})),p=d(()=>a(s).annotation);let e=d(()=>n().data);const L=d(qt),x=d(()=>a(L).taskMap),$=d(()=>a(e)&&a(e).annotation_task_id&&V(a(x)).has(a(e).annotation_task_id)?V(a(x)).get(a(e).annotation_task_id).is_prediction:!1),I=d(()=>{var l,u;if(!a(e))return[];let c=[{id:"label",label:"Label:",value:(u=(l=a(e))==null?void 0:l.annotation_label)==null?void 0:u.annotation_label_name},{id:"type",label:"Type:",value:Aa[a(e).annotation_type]||"Unknown"},{id:"status",label:"Status:",value:a($)?"Prediction":"Ground Truth"}];if(a(e)&&!Gt(a(e))){const{height:h,width:M}=xt(a(e));c=[{id:"height",label:"Height:",value:ht(Math.round(h))+"px"},{id:"width",label:"Width:",value:ht(Math.round(M))+"px"},...c]}return c}),{isEditingMode:g}=z.data.globalStorage;var P=Z(),y=B(P);{var A=c=>{Kt(c,{title:"Annotation details",children:(l,u)=>{var h=Pa(),M=f(h);Wt(M,21,()=>a(I),({label:N,value:b,id:C})=>N,(N,b)=>{let C=()=>a(b).label,E=()=>a(b).value,H=()=>a(b).id;var G=Z(),nt=B(G);{var Q=w=>{ka(w,{get onUpdate(){return t.onUpdate},get annotationId(){return t.annotationId},datasetId:o,get label(){return C()},get value(){return E()},isEditingMode:g})},et=w=>{Sa(w,{get label(){return C()},get id(){return H()},get value(){return E()}})};O(nt,w=>{H()==="label"&&g?w(Q):w(et,!1)})}_(N,G)}),m(M),m(h),_(l,h)}})};O(y,c=>{a(I).length>0&&c(A)})}_(v,P),j(),i()}var Na=k('<!> <h2 class="py-2 text-lg font-semibold"><a>Check all sample annotations</a></h2>',1),Ca=k('<div class="flex h-full w-full items-center justify-center"><!></div>'),Ba=k('<div class="flex h-full min-h-0 flex-col space-y-4 overflow-hidden dark:[color-scheme:dark]"><!> <!></div>');function La(v,t){T(t,!0);const{datasetId:r}=z.data;let i=Tt(void 0);at({datasetId:r,annotationId:t.annotationId}).annotation.subscribe(e=>{e.isSuccess&&Dt(i,fa(e.data.sample))});const o=Vt(),[s,p]=o;It(v,{className:"h-full",children:(e,L)=>{At(e,{className:"h-full flex flex-col",children:(x,$)=>{var I=Ba(),g=f(I);Ma(g,{get annotationId(){return t.annotationId},get onUpdate(){return t.onUpdate}});var P=S(g,2);{var y=c=>{var l=Na(),u=B(l);na(u,{get sample(){return a(i)},showCustomMetadata:!1});var h=S(u,2),M=f(h);m(h),F(N=>J(M,"href",N),[()=>s(U.toSample({sampleId:a(i).sample_id,datasetId:a(i).dataset_id}),void 0)]),_(c,l)},A=c=>{var l=Ca(),u=f(l);wt(u,{size:"large",align:"center"}),m(l),_(c,l)};O(P,c=>{a(i)?c(y):c(A,!1)})}m(I),_(x,I)},$$slots:{default:!0}})},$$slots:{default:!0}}),j()}var Ea=k('<!> <span class="hidden sm:inline">Home</span>',1),Da=k('<!> <span class="max-w-[150px] truncate"> </span>',1),Ta=k('<!> <span class="hidden sm:inline">Annotations</span>',1),ja=k('<!> <span class="max-w-[200px] truncate"> </span>',1),Ha=k("<!> <!> <!> <!> <!> <!> <!>",1);function Ra(v,t){T(t,!0);const[r,i]=q(),n=()=>D(o,"$annotationsTotalCount",r),{annotationsTotalCount:o}=St();ea(v,{class:"mb-2",children:(s,p)=>{oa(s,{children:(e,L)=>{var x=Ha(),$=B(x);X($,{children:(l,u)=>{const h=d(()=>U.toHome());lt(l,{get href(){return a(h)},class:"flex items-center gap-2",children:(M,N)=>{var b=Ea(),C=B(b);sa(C,{class:"h-4 w-4"}),gt(2),_(M,b)},$$slots:{default:!0}})},$$slots:{default:!0}});var I=S($,2);dt(I,{});var g=S(I,2);X(g,{children:(l,u)=>{const h=d(()=>U.toAnnotations(t.dataset.dataset_id));lt(l,{get href(){return a(h)},class:"flex items-center gap-2",get title(){return t.dataset.directory},children:(M,N)=>{var b=Da(),C=B(b);ra(C,{class:"h-4 w-4"});var E=S(C,2),H=f(E,!0);m(E),F(()=>{J(E,"title",t.dataset.directory),W(H,t.dataset.name)}),_(M,b)},$$slots:{default:!0}})},$$slots:{default:!0}});var P=S(g,2);dt(P,{});var y=S(P,2);X(y,{children:(l,u)=>{const h=d(()=>U.toAnnotations(t.dataset.dataset_id));lt(l,{get href(){return a(h)},class:"flex items-center gap-2",children:(M,N)=>{var b=Ta(),C=B(b);_a(C,{class:"h-4 w-4"}),gt(2),_(M,b)},$$slots:{default:!0}})},$$slots:{default:!0}});var A=S(y,2);dt(A,{});var c=S(A,2);X(c,{children:(l,u)=>{ia(l,{class:"flex items-center gap-2",children:(h,M)=>{var N=ja(),b=B(N);ga(b,{class:"h-4 w-4"});var C=S(b,2),E=f(C);m(C),F(()=>W(E,`Annotation ${t.annotationIndex+1} of ${n()??""}`)),_(h,N)},$$slots:{default:!0}})},$$slots:{default:!0}}),_(e,x)},$$slots:{default:!0}})},$$slots:{default:!0}}),j(),i()}var Ua=Ht("<image></image><g><!></g>",1),Fa=k('<div class="h-full w-full overflow-hidden"><div class="sample relative h-full w-full"><div class="absolute right-4 top-2 z-30"><!></div> <!> <!></div></div>'),Oa=k('<div class="flex h-full w-full items-center justify-center"><!></div>'),za=k('<div class="flex h-full w-full flex-col space-y-4"><div class="flex w-full items-center"><!></div> <!> <div class="flex min-h-0 flex-1 gap-4"><div class="flex-1"><!></div> <div class="relative w-[375px]"><!></div></div></div>');function Ga(v,t){T(t,!0);const[r,i]=q(),n=()=>D(a(l),"$annotationResp",r),o=()=>D(e,"$selectedSampleAnnotationCropIds",r),s=()=>D(L,"$isHidden",r),{toggleSampleAnnotationCropSelection:p,selectedSampleAnnotationCropIds:e}=St(),{isHidden:L,handleKeyEvent:x}=Zt(),{settingsStore:$}=Jt(),I=" ",g=V($).key_go_back,P=()=>{var w;(w=t.sample)!=null&&w.dataset_id?tt(U.toAnnotations(t.sample.dataset_id)):tt("/")},y=w=>{switch(w.key){case g:P();break;case I:w.preventDefault(),p(t.annotationId);break}x(w)},A=z.data.datasetId,c=d(()=>at({datasetId:A,annotationId:t.annotationId})),l=d(()=>a(c).annotation);let u=d(()=>n().data),h=d(()=>{var w,ot;return da(((ot=(w=a(u))==null?void 0:w.sample)==null?void 0:ot.sample_id)||"")}),M=d(()=>a(u)?xt(a(u)):void 0);var N=za();ut("keydown",ct,y),ut("keyup",ct,x);var b=f(N),C=f(b);Ra(C,{get dataset(){return t.dataset},get annotationIndex(){return t.annotationIndex}}),m(b);var E=S(b,2);ta(E,{class:"mb-4 bg-border-hard"});var H=S(E,2),G=f(H),nt=f(G);It(nt,{className:"h-full",children:(w,ot)=>{At(w,{className:"h-full",children:($t,Wa)=>{var vt=Z(),yt=B(vt);{var kt=K=>{var R=Fa(),Y=f(R),st=f(Y),Mt=f(st);const Nt=d(()=>o().has(t.annotationId));Qt(Mt,{onSelect:()=>p(t.annotationId),get isSelected(){return a(Nt)}}),m(st);var mt=S(st,2);Ia(mt,{});var Ct=S(mt,2);la(Ct,{get width(){return a(u).sample.width},get height(){return a(u).sample.height},get boundingBox(){return a(M)},children:(Bt,qa)=>{var ft=Ua(),_t=B(ft),rt=S(_t);let pt;var Lt=f(rt);Yt(Lt,()=>a(u).annotation_id,it=>{aa(it,{get annotation(){return a(u)},showLabel:!0,get imageWidth(){return a(u).sample.width}})}),m(rt),F(it=>{J(_t,"href",a(h)),pt=Xt(rt,0,"",null,pt,it)},[()=>({invisible:s()})]),_(Bt,ft)},$$slots:{default:!0}}),m(Y),m(R),_(K,R)},Pt=K=>{var R=Oa(),Y=f(R);wt(Y,{size:"large",align:"center"}),m(R),_(K,R)};O(yt,K=>{a(u)?K(kt):K(Pt,!1)})}_($t,vt)},$$slots:{default:!0}})},$$slots:{default:!0}}),m(G);var Q=S(G,2),et=f(Q);La(et,{get annotationId(){return t.annotationId}}),m(Q),m(H),m(N),_(v,N),j(),i()}var Ka=k('<div class="flex h-full w-full space-x-4 px-4 pb-4" data-testid="annotation-details"><div class="h-full w-full space-y-6 rounded-[1vw] bg-card p-4"><!></div></div>');function hn(v,t){T(t,!0);const[r,i]=q(),n=()=>D(a(x),"$sample",r),o=d(()=>t.data.annotationId),s=d(()=>t.data.dataset),p=d(()=>t.data.annotationIndex),{sampleId:e}=z.params,L=d(()=>ca({sampleId:e,datasetId:a(s).dataset_id})),x=d(()=>a(L).sample);var $=Ka(),I=f($),g=f(I);{var P=y=>{Ga(y,{get annotationId(){return a(o)},get annotationIndex(){return a(p)},get dataset(){return a(s)},get sample(){return n().data}})};O(g,y=>{n().data&&a(o)&&a(s)&&y(P)})}m(I),m($),_(v,$),j(),i()}export{hn as component,gn as universal};

lightly_studio-0.3.1.dist-info/METADATA

@@ -1,520 +0,0 @@
-Metadata-Version: 2.4
-Name: lightly-studio
-Version: 0.3.1
-Summary: LightlyStudio is a lightweight, fast, and easy-to-use data exploration tool for data scientists and engineers.
-Classifier: Operating System :: MacOS :: MacOS X
-Classifier: Operating System :: Microsoft :: Windows
-Classifier: Operating System :: POSIX :: Linux
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.8
-Classifier: Programming Language :: Python :: 3.9
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Requires-Python: >=3.8
-Requires-Dist: annotated-types==0.7.0
-Requires-Dist: duckdb-engine>=0.13.5
-Requires-Dist: duckdb>=1.1.3
-Requires-Dist: environs<12.0.0
-Requires-Dist: eval-type-backport>=0.2.2
-Requires-Dist: fastapi>=0.115.5
-Requires-Dist: faster-coco-eval>=1.6.5
-Requires-Dist: labelformat>=0.1.7
-Requires-Dist: lightly-mundig==0.1.3
-Requires-Dist: open-clip-torch>=2.20.0
-Requires-Dist: python-multipart>=0.0.20
-Requires-Dist: scikit-learn==1.3.2
-Requires-Dist: sqlmodel>=0.0.22
-Requires-Dist: torchmetrics>=1.5.2
-Requires-Dist: tqdm>=4.65.0
-Requires-Dist: typing-extensions>=4.12.2
-Requires-Dist: uvicorn>=0.32.1
-Requires-Dist: xxhash>=3.5.0
-Description-Content-Type: text/markdown
-
-<div align="center">
-<p align="center">
-
-<!-- prettier-ignore -->
-<img src="https://cdn.prod.website-files.com/62cd5ce03261cba217188442/66dac501a8e9a90495970876_Logo%20dark-short-p-800.png" height="50px">
-
-**The open-source tool curating datasets**
-
----
-
-[![PyPI python](https://img.shields.io/pypi/pyversions/lightly-studio)](https://pypi.org/project/lightly-studio)
-[![PyPI version](https://badge.fury.io/py/lightly-studio.svg)](https://pypi.org/project/lightly-studio)
-[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](LICENSE)
-
-</p>
-</div>
-
-# 🚀 Welcome to LightlyStudio!
-
-We at **[Lightly](https://lightly.ai)** created **LightlyStudio**, an open-source tool
-designed to supercharge your data curation workflows for computer vision datasets. Explore
-your data, visualize annotations and crops, tag samples, and export curated lists to improve
-your machine learning pipelines. And much more!
-
-LightlyStudio runs entirely locally on your machine, keeping your data private. It consists
-of a Python library for indexing your data and a web-based UI for visualization and curation.
-
-## ✨ Core Workflow
-
-Using LightlyStudio typically involves these steps:
-
-1.  **Index Your Dataset:** Run a Python script using the `lightly_studio` library to process your local dataset (images and annotations) and save metadata into a local `lightly_studio.db` file.
-2.  **Launch the UI:** The script then starts a local web server.
-3.  **Explore & Curate:** Use the UI to visualize images, annotations, and object crops. Filter and search your data (experimental text search available). Apply tags to interesting samples (e.g., "mislabeled", "review").
-4.  **Export Curated Data:** Export information (like filenames) for your tagged samples from the UI to use downstream.
-5.  **Stop the Server:** Close the terminal running the script (Ctrl+C) when done.
-
-<p align="center">
-  <img alt="LightlyStudio Sample Grid View" src="https://storage.googleapis.com/lightly-public/studio/screenshot_grid_view.jpg" width="70%">
-  <br/>
-  <em>Visualize your dataset samples with annotations in the grid view.</em>
-</p>
-<p align="center">
-  <img alt="LightlyStudio Annotation Crop View" src="https://storage.googleapis.com/lightly-public/studio/screenshot_annotation_view.jpg" width="70%">
-  <br/>
-  <em>Switch to the annotation view to inspect individual object crops easily.</em>
-</p>
-<p align="center">
-  <img alt="LightlyStudio Sample Detail View" src="https://storage.googleapis.com/lightly-public/studio/screenshot_detail_view.jpg" width="70%">
-  <br/>
-  <em>Inspect individual samples in detail, viewing all annotations and metadata.</em>
-</p>
-
-## 🎯 Features
-
-* **Local Web GUI:** Explore and curate your dataset in your browser. Works completelly
-  offline, your data never leaves your machine.
-* **Flexible Input Formats:** Load your image dataset from a folder, or with annotations from
-  a number of popular formats like e.g. COCO or YOLO.
-* **Metadata:** Attach your custom metadata to every sample.
-* **Tags:** Mark subsets of your dataset for later use.
-* **Embeddings:** Run similarity search queries on your data.
-* **Selection:** Run advanced selection algorithms to tag a subset of your data.
-
-## 💻 Installation
-
-Ensure you have **Python 3.8 or higher**. We strongly recommend using a virtual environment.
-
-The library is OS-independent and works on Windows, Linux, and macOS.
-
-```shell
-# 1. Create and activate a virtual environment (Recommended)
-# On Linux/macOS:
-python3 -m venv venv
-source venv/bin/activate
-
-# On Windows:
-python -m venv venv
-.\venv\Scripts\activate
-
-# 2. Install LightlyStudio
-pip install lightly_studio
-```
-
-## **Quickstart**
-
-Download the dataset and run a quickstart script to load your dataset and launch the app.
-
-### YOLO Object Detection
-
-To run an example using a yolo dataset, clone the example repository and run the example script:
-
-```shell
-git clone https://github.com/lightly-ai/dataset_examples_studio dataset_examples_studio
-python dataset_examples_studio/road_signs_yolo/example_yolo.py
-```
-
-<details>
-<summary>The YOLO format details:</summary>
-
-```
-road_signs_yolo/
-├── train/
-│   ├── images/
-│   │   ├── image1.jpg
-│   │   ├── image2.jpg
-│   │   └── ...
-│   └── labels/
-│       ├── image1.txt
-│       ├── image2.txt
-│       └── ...
-├── valid/  (optional)
-│   ├── images/
-│   │   └── ...
-│   └── labels/
-│       └── ...
-└── data.yaml
-```
-
-Each label file should contain YOLO format annotations (one per line):
-
-```
-<class> <x_center> <y_center> <width> <height>
-```
-
-Where coordinates are normalized between 0 and 1.
-
-</details>
-
-
-<details>
-<summary>Let's break down the `example_yolo.py` script to explore the dataset:</summary>
-
-```python
-# We import the DatasetLoader class from the lightly_studio module
-from lightly_studio import DatasetLoader
-from pathlib import Path
-
-# Create a DatasetLoader instance
-loader = DatasetLoader()
-
-data_yaml_path = Path(__file__).resolve().parent / "data.yaml"
-loader.from_yolo(
-    data_yaml_path=str(data_yaml_path),
-    input_split="test",
-)
-
-# We start the UI application on the port defined with the LIGHTLY_STUDIO_PORT env variable or 8001 by default.
-loader.start_gui()
-```
-</details>
-
-</details>
-
-### COCO Instance Segmentation
-
-To run an instance segmentation example using a COCO dataset, clone the example repository and run the example script:
-
-```shell
-git clone https://github.com/lightly-ai/dataset_examples_studio dataset_examples_studio
-python dataset_examples_studio/coco_subset_128_images/example_coco.py
-```
-
-<details>
-<summary>The COCO format details:</summary>
-
-```
-coco_subset_128_images/
-├── images/
-│   ├── image1.jpg
-│   ├── image2.jpg
-│   └── ...
-└── instances_train2017.json        # Single JSON file containing all annotations
-```
-
-COCO uses a single JSON file containing all annotations. The format consists of three main components:
-
-- Images: Defines metadata for each image in the dataset.
-- Categories: Defines the object classes.
-- Annotations: Defines object instances.
-
-</details>
-
-<details>
-<summary>Let's break down the `example_coco.py` script to explore the dataset:</summary>
-
-```python
-# We import the DatasetLoader class from the lightly_studio module
-from lightly_studio import DatasetLoader
-from pathlib import Path
-
-# Create a DatasetLoader instance
-loader = DatasetLoader()
-
-current_dir = Path(__file__).resolve().parent
-loader.from_coco_instance_segmentations(
-    annotations_json_path=str(current_dir / "instances_train2017.json"),
-    img_dir=str(current_dir / "images"),
-)
-
-# We start the UI application on the port defined with the LIGHTLY_STUDIO_PORT env variable or 8001 by default.
-loader.start_gui()
-
-```
-</details>
-
-</details>
-
-## 🔍 How It Works
-
-1.  Your **Python script** uses the `lightly_studio` **Dataset Loader**.
-2.  The Loader reads your images and annotations, calculates embeddings, and saves metadata to a local **`lightly_studio.db`** file (using DuckDB).
-3.  `loader.start_gui()` starts a **local Backend API** server.
-4.  This server reads from `lightly_studio.db` and serves data to the **UI Application** running in your browser (`http://localhost:8001`).
-5.  Images are streamed directly from your disk for display in the UI.
-
-## 🎯 Python Interface
-
-### Load Image Dataset From A Folder
-
-```py
-from lightly_studio import DatasetLoader
-
-loader = DatasetLoader()
-loader.from_directory(
-    dataset_name="my-dataset",
-    img_dir="/path/to/image_dataset",
-)
-loader.start_gui()
-```
-
-### Load Image Dataset With Annotations
-
-The `DatasetLoader` currently supports:
-
-- **YOLOv8 Object Detection:** Reads `.yaml` file. Supports bounding boxes.
-- **COCO Object Detection:** Reads `.json` annotations. Supports bounding boxes.
-- **COCO Instance Segmentation:** Reads `.json` annotations. Supports instance masks in RLE (Run-Length Encoding) format.
-
-```py
-from lightly_studio import DatasetLoader
-
-# Create a DatasetLoader instance
-loader = DatasetLoader()
-
-# Load a dataset in YOLO format
-dataset = loader.from_yolo(
-    data_yaml_path="my_yolo_dataset/data.yaml",
-    input_split="test",
-)
-
-# Load an object detection dataset in COCO format
-dataset = loader.from_coco_object_detections(
-    annotations_json_path="my_coco_dataset/detections_train.json",
-    img_dir="my_coco_dataset/images",
-)
-
-# Load an instance segmentation dataset in COCO format
-dataset = loader.from_coco_instance_segmentations(
-    annotations_json_path="my_coco_dataset/instances_train.json",
-    img_dir="my_coco_dataset/images",
-)
-
-# Launch the GUI
-loader.start_gui()
-```
-
-### Samples
-
-The dataset consists of samples. Every sample corresponds to an image.
-Dataset samples can be fetched and accessed as follows,
-for a full list of attributes see [src/lightly_studio/models/sample.py]().
-
-```py
-# Get all dataset samples
-samples = dataset.get_samples()
-
-# Access sample attributes
-s = samples[0]
-s.sample_id        # Sample ID
-s.file_name        # Image file name
-s.file_path_abs    # Full image file path
-s.tags             # The list of sample tags
-...
-```
-
-### Sample Filtering
-
-You can efficiently fetch filtered dataset samples with the `get_samples()` method.
-All arguments are optional. A database object UUID is needed for filtering by
-annotation labels, tags and samples. They would be obtained by other function calls.
-
-```py
-from lightly_studio.resolvers.samples_filter import FilterDimensions, SampleFilter
-from lightly_studio.resolvers.metadata_resolver.metadata_filter import Metadata
-
-sample1 = ...
-sample2 = ...
-ann1 = ...
-tag1 = ...
-
-# Example with available filters, all arguments are optional.
-samples = dataset.get_samples(
-    # Set offset and limit to the list of returned samples.
-    # Samples are ordered by their creation date.
-    offset=0,
-    limit=10,
-
-    filters=SampleFilter(
-        # Filter by width and/or height.
-        width=FilterDimensions(min=10, max=200),
-        height=FilterDimensions(min=10, max=None),
-
-        # Filter by annotations.
-        annotation_label_ids=[ann1.annotation_label_id],
-
-        # Filter by tags.
-        tag_ids=[tag1.tag_id],
-
-        # Filter by metadata.
-        metadata_filters=[
-            Metadata("temperature") > 25.0,
-            Metadata("nested.bool_key") == True,
-        ]
-    ),
-
-    # Fetch only specific samples by IDs.
-    sample_ids=[sample1.sample_id, sample2.sample_id],
-)
-```
-
-Alternatively, samples can be filtered directly in python code:
-
-```py
-my_samples = [
-    sample in dataset.get_samples()
-    if sample.file_name in ["image1.jpg", "image2.jpg"]
-]
-```
-
-### Add Custom Metadata
-
-Attach values to custom fields for every sample.
-
-```py
-from lightly_studio import DatasetLoader
-
-# Load your dataset
-loader = DatasetLoader()
-dataset = loader.from_directory(
-    dataset_name="my-dataset",
-    img_dir="/path/to/image_dataset",
-)
-
-# Attach metadata
-for sample in dataset.get_samples():
-    sample["my_metadata"] = f"Example metadata field for {sample.file_name}"
-    sample["my_dict"] = {"my_int_key": 10, "my_bool_key": True}
-
-# View metadata in GUI
-loader.start_gui()
-```
-
-### Tags
-
-You can easily mark subsets of your data with tags.
-
-```py
-from lightly_studio.resolvers import tag_resolver
-from lightly_studio.models.tag import TagCreate
-
-# Load your dataset
-loader = DatasetLoader()
-dataset = ...
-
-# Create a tag
-my_tag = tag_resolver.create(
-    session=loader.session,
-    tag=TagCreate(dataset_id=dataset.dataset_id, name="my-tag"),
-)
-
-# Tag some samples, for example the first 10 samples:
-for sample in dataset.get_samples()[:10]:
-    tag_resolver.add_tag_to_sample(
-        session=loader.session,
-        tag_id=my_tag.tag_id,
-        sample=sample,
-    )
-```
-
-Find existing tags and tagged samples as follows.
-
-```py
-from lightly_studio.resolvers import tag_resolver
-from lightly_studio.resolvers.samples_filter import SampleFilter
-
-# Get all tags
-all_tags = tag_resolver.get_all_by_dataset_id(
-    session=loader.session,
-    dataset_id=dataset.dataset_id,
-)
-
-# Get a tag by name
-my_tag = tag_resolver.get_by_name(
-    session=loader.session,
-    tag_name="my-tag"
-)
-
-# Get tagged samples
-tagged_samples = dataset.get_samples(filters=SampleFilter(tag_ids=[my_tag.tag_id]))
-```
-
-### Selection
-
-LightlyStudio offers as a premium feature advanced methods for subselecting dataset
-samples.
-
-**Prerequisites:** The selection functionality requires a valid LightlyStudio license key.
-Set the `LIGHTLY_STUDIO_LICENSE_KEY` environment variable before using selection features:
-
-```bash
-export LIGHTLY_STUDIO_LICENSE_KEY="license_key_here"
-```
-Alternatively, set it inside your Python script:
-
-```py
-import os
-os.environ["LIGHTLY_STUDIO_LICENSE_KEY"] = "license_key_here"
-```
-
-The selection can be configured with a rich interface of the `Selection` object. The
-example below showcases a simple case of selecting diverse samples.
-
-```py
-import os
-from lightly_studio import DatasetLoader
-from lightly_studio.selection.select import Selection
-
-# Load your dataset
-loader = DatasetLoader()
-dataset = loader.from_directory(
-    dataset_name="my-dataset",
-    img_dir="/path/to/image_dataset",
-)
-
-# Select a diverse subset of 10 samples.
-select = Selection(dataset_id=dataset.dataset_id, session=loader.session)
-select.diverse(
-    n_samples_to_select=10,
-    selection_result_tag_name="diverse_selection",
-)
-
-loader.start_gui()
-```
-
-The selected sample paths can be exported via the GUI, or by a script:
-
-```py
-from lightly_studio.resolvers import tag_resolver
-from lightly_studio.resolvers.samples_filter import SampleFilter
-
-tag = tag_resolver.get_by_name(session=loader.session, tag_name="diverse_selection")
-selected_samples = dataset.get_samples(filters=SampleFilter(tag_ids=[tag.tag_id]))
-
-with open("export.txt", "w") as f:
-    for sample in selected_samples:
-        f.write(f"{sample.file_name}\n")
-```
-
-## 📚 **FAQ**
-
-### Are the datasets persistent?
-
-Yes, the information about datasets is persistent and stored in the db file. You can see it after the dataset is processed.
-If you rerun the loader it will create a new dataset representing the same dataset, keeping the previous dataset information untouched.
-
-### Can I change the database path?
-
-Not yet. The database is stored in the working directory by default.
-
-### Can I launch in another Python script or do I have to do it in the same script?
-
-It is possible to use only one script at the same time because we lock the db file for the duration of the script.
-
-### Can I change the API backend port?
-
-Yes. To change the port set the LIGHTLY_STUDIO_PORT variable to your preffered value. If at runtime the port is unavailable it will try to set it to a random value.