Upload folder using huggingface_hub

Charades/Charades_v1_classes.txt

@@ -0,0 +1,157 @@
+c000 Holding some clothes
+c001 Putting clothes somewhere
+c002 Taking some clothes from somewhere
+c003 Throwing clothes somewhere
+c004 Tidying some clothes
+c005 Washing some clothes
+c006 Closing a door
+c007 Fixing a door
+c008 Opening a door
+c009 Putting something on a table
+c010 Sitting on a table
+c011 Sitting at a table
+c012 Tidying up a table
+c013 Washing a table
+c014 Working at a table
+c015 Holding a phone/camera
+c016 Playing with a phone/camera
+c017 Putting a phone/camera somewhere
+c018 Taking a phone/camera from somewhere
+c019 Talking on a phone/camera
+c020 Holding a bag
+c021 Opening a bag
+c022 Putting a bag somewhere
+c023 Taking a bag from somewhere
+c024 Throwing a bag somewhere
+c025 Closing a book
+c026 Holding a book
+c027 Opening a book
+c028 Putting a book somewhere
+c029 Smiling at a book
+c030 Taking a book from somewhere
+c031 Throwing a book somewhere
+c032 Watching/Reading/Looking at a book
+c033 Holding a towel/s
+c034 Putting a towel/s somewhere
+c035 Taking a towel/s from somewhere
+c036 Throwing a towel/s somewhere
+c037 Tidying up a towel/s
+c038 Washing something with a towel
+c039 Closing a box
+c040 Holding a box
+c041 Opening a box
+c042 Putting a box somewhere
+c043 Taking a box from somewhere
+c044 Taking something from a box
+c045 Throwing a box somewhere
+c046 Closing a laptop
+c047 Holding a laptop
+c048 Opening a laptop
+c049 Putting a laptop somewhere
+c050 Taking a laptop from somewhere
+c051 Watching a laptop or something on a laptop
+c052 Working/Playing on a laptop
+c053 Holding a shoe/shoes
+c054 Putting shoes somewhere
+c055 Putting on shoe/shoes
+c056 Taking shoes from somewhere
+c057 Taking off some shoes
+c058 Throwing shoes somewhere
+c059 Sitting in a chair
+c060 Standing on a chair
+c061 Holding some food
+c062 Putting some food somewhere
+c063 Taking food from somewhere
+c064 Throwing food somewhere
+c065 Eating a sandwich
+c066 Making a sandwich
+c067 Holding a sandwich
+c068 Putting a sandwich somewhere
+c069 Taking a sandwich from somewhere
+c070 Holding a blanket
+c071 Putting a blanket somewhere
+c072 Snuggling with a blanket
+c073 Taking a blanket from somewhere
+c074 Throwing a blanket somewhere
+c075 Tidying up a blanket/s
+c076 Holding a pillow
+c077 Putting a pillow somewhere
+c078 Snuggling with a pillow
+c079 Taking a pillow from somewhere
+c080 Throwing a pillow somewhere
+c081 Putting something on a shelf
+c082 Tidying a shelf or something on a shelf
+c083 Reaching for and grabbing a picture
+c084 Holding a picture
+c085 Laughing at a picture
+c086 Putting a picture somewhere
+c087 Taking a picture of something
+c088 Watching/looking at a picture
+c089 Closing a window
+c090 Opening a window
+c091 Washing a window
+c092 Watching/Looking outside of a window
+c093 Holding a mirror
+c094 Smiling in a mirror
+c095 Washing a mirror
+c096 Watching something/someone/themselves in a mirror
+c097 Walking through a doorway
+c098 Holding a broom
+c099 Putting a broom somewhere
+c100 Taking a broom from somewhere
+c101 Throwing a broom somewhere
+c102 Tidying up with a broom
+c103 Fixing a light
+c104 Turning on a light
+c105 Turning off a light
+c106 Drinking from a cup/glass/bottle
+c107 Holding a cup/glass/bottle of something
+c108 Pouring something into a cup/glass/bottle
+c109 Putting a cup/glass/bottle somewhere
+c110 Taking a cup/glass/bottle from somewhere
+c111 Washing a cup/glass/bottle
+c112 Closing a closet/cabinet
+c113 Opening a closet/cabinet
+c114 Tidying up a closet/cabinet
+c115 Someone is holding a paper/notebook
+c116 Putting their paper/notebook somewhere
+c117 Taking paper/notebook from somewhere
+c118 Holding a dish
+c119 Putting a dish/es somewhere
+c120 Taking a dish/es from somewhere
+c121 Wash a dish/dishes
+c122 Lying on a sofa/couch
+c123 Sitting on sofa/couch
+c124 Lying on the floor
+c125 Sitting on the floor
+c126 Throwing something on the floor
+c127 Tidying something on the floor
+c128 Holding some medicine
+c129 Taking/consuming some medicine
+c130 Putting groceries somewhere
+c131 Laughing at television
+c132 Watching television
+c133 Someone is awakening in bed
+c134 Lying on a bed
+c135 Sitting in a bed
+c136 Fixing a vacuum
+c137 Holding a vacuum
+c138 Taking a vacuum from somewhere
+c139 Washing their hands
+c140 Fixing a doorknob
+c141 Grasping onto a doorknob
+c142 Closing a refrigerator
+c143 Opening a refrigerator
+c144 Fixing their hair
+c145 Working on paper/notebook
+c146 Someone is awakening somewhere
+c147 Someone is cooking something
+c148 Someone is dressing
+c149 Someone is laughing
+c150 Someone is running somewhere
+c151 Someone is going from standing to sitting
+c152 Someone is smiling
+c153 Someone is sneezing
+c154 Someone is standing up from somewhere
+c155 Someone is undressing
+c156 Someone is eating something

Charades/Charades_v1_classify.m

@@ -0,0 +1,128 @@
+function [rec_all,prec_all,ap_all,map]=Charades_v1_classify(clsfilename,gtpath)
+%
+%     Input:    clsfilename: path of the input file
+%                    gtpath: the path of the groundtruth file
+%
+%    Output:        rec_all: recall
+%                  prec_all: precision
+%                    ap_all: AP for each class
+%                       map: MAP 
+%
+% Example:
+%
+%  [rec_all,prec_all,ap_all,map]=Charades_v1_classify('test_submission_classify.txt','Charades_v1_test.csv');
+%
+% Code adapted from THUMOS15 
+%
+
+[gtids,gtclasses] = load_charades(gtpath);
+nclasses = 157;
+ntest = length(gtids);
+
+% load test scores
+[testids,testscores]=textread(clsfilename,'%s%[^\n]');
+nInputNum=size(testscores,1);
+if nInputNum<ntest
+    fprintf('Warning: %d Videos missing\n',ntest-nInputNum);
+end
+for i=1:nInputNum
+    id = testids{i};
+    z=regexp(testscores{i},'\t','split');
+    eleNum=size(z,2);
+    if eleNum~=nclasses&&eleNum~=nclasses+1
+        z=regexp(testscores{i},' ','split');
+    end
+    eleNum=size(z,2);
+    if eleNum~=nclasses&&eleNum~=nclasses+1
+        fprintf('Error: Incompatible number of classes\n');
+    end
+    for j=1:eleNum
+        z{j}=regexprep(z{j},'\t','');
+        z{j}=regexprep(z{j},' ','');
+    end
+    x = zeros(nclasses,1);
+    for j=1:nclasses
+	x(j) = str2double(z{j});
+    end
+    testscores{i} = x;
+end
+predictions = containers.Map(testids,testscores);
+
+% compare test scores to ground truth
+gtlabel = zeros(ntest,nclasses);
+test = -inf(ntest,nclasses);
+for i=1:ntest
+    id = gtids{i};
+    gtlabel(i,gtclasses{i}+1) = 1;
+    if predictions.isKey(id)
+	test(i,:) = predictions(id);
+    end
+end
+
+for i=1:nclasses
+    [rec_all(:,i),prec_all(:,i),ap_all(:,i)]=THUMOSeventclspr(test(:,i),gtlabel(:,i));
+end
+map=mean(ap_all);
+wap=sum(ap_all.*sum(gtlabel,1))/sum(gtlabel(:));
+fprintf('\n\n')
+fprintf('MAP: %f\n',map);
+fprintf('WAP: %f (weighted by size of each class)',wap);
+fprintf('\n\n')
+
+
+function [rec,prec,ap]=THUMOSeventclspr(conf,labels)
+[so,sortind]=sort(-conf);
+tp=labels(sortind)==1;
+fp=labels(sortind)~=1;
+npos=length(find(labels==1));
+
+% compute precision/recall
+fp=cumsum(fp);
+tp=cumsum(tp);
+rec=tp/npos;
+prec=tp./(fp+tp);
+
+% compute average precision
+
+ap=0;
+tmp=labels(sortind)==1;
+for i=1:length(conf)
+    if tmp(i)==1
+        ap=ap+prec(i);
+    end
+end
+ap=ap/npos;
+
+
+function [gtids,gtclasses] = load_charades(gtpath)
+f = fopen(gtpath);
+
+% read column headers
+headerline = textscan(f,'%s',1);
+headerline = regexp(headerline{1}{1},',','split');
+ncols = length(headerline);
+headers = struct();
+for i=1:ncols
+    headers = setfield(headers,headerline{i},i);
+end
+
+% read data
+gtcsv = textscan(f,repmat('%q ',[1 ncols]),'Delimiter',',');
+ntest = size(gtcsv{1},1);
+gtids = cell(ntest,1);
+gtclasses = cell(ntest,1);
+for i=1:ntest
+    id = gtcsv{headers.id}{i};
+    classes = gtcsv{headers.actions}{i};
+    if length(classes)==0; gtclasses{i} = []; continue; end
+    classes = regexp(classes,';','split');
+    for j=1:length(classes)
+	tmp = regexp(classes{j},' ','split');
+	[class,s,e] = tmp{:};
+	classes{j} = str2double(class(2:end));
+    end
+    gtids{i} = id;
+    gtclasses{i} = cell2mat(classes);
+end
+
+

Charades/Charades_v1_localize.m

@@ -0,0 +1,155 @@
+function [rec_all,prec_all,ap_all,map]=Charades_v1_localize(clsfilename,gtpath)
+%
+%     Input:    clsfilename: path of the input file
+%                    gtpath: the path of the groundtruth file
+%
+%    Output:        rec_all: recall
+%                  prec_all: precision
+%                    ap_all: AP for each class
+%                       map: MAP 
+%
+% Please refer to the README.txt file for an overview of how localization performance is evaluated
+% 
+% Example:
+%
+%  [rec_all,prec_all,ap_all,map]=Charades_v1_localize('test_submission.txt','Charades_v1_test.csv');
+%
+% Code adapted from THUMOS15 
+%
+
+tic;
+fprintf('Loading Charades Annotations:\n');
+frames_per_video = 25;
+[gtids,gtclasses] = load_charades_localized(gtpath,frames_per_video);
+nclasses = 157;
+ntest = length(gtids);
+toc; tic;
+
+% load test scores
+fprintf('Reading Submission File:\n');
+[testids,framenr,testscores]=textread(clsfilename,'%s%d%[^\n]');
+if min(framenr)==0 
+    fprintf('Warning: Frames should be 1 indexed\n');
+    fprintf('Warning: Adding 1 to all frames numbers\n');
+    framenr = framenr+1;
+end
+toc; tic;
+fprintf('Parsing Submission Scores:\n'); 
+nInputNum=size(testscores,1);
+if nInputNum<ntest
+    fprintf('Warning: %d Total frames missing\n',ntest-nInputNum);
+end
+testscoresparsed = cellfun(@str2num,testscores,'UniformOutput',false);
+eleNum=length(testscoresparsed{1});
+if eleNum~=nclasses&&eleNum~=nclasses+1
+    fprintf('Error: Incompatible number of classes\n');
+end
+make_frameid = @(x,y) [x,'-',sprintf('%03d',y)];
+frameids = cellfun(make_frameid,testids,num2cell(framenr),'UniformOutput',false);
+predictions = containers.Map(frameids,testscoresparsed);
+toc; tic;
+
+% compare test scores to ground truth
+fprintf('Constructing Ground Truth Matrix:\n')
+gtlabel = zeros(ntest,nclasses);
+test = -inf(ntest,nclasses);
+for i=1:ntest
+    id = gtids{i};
+    gtlabel(i,gtclasses{i}+1) = 1;
+    if predictions.isKey(id)
+        test(i,:) = predictions(id);
+    end
+end
+toc; tic;
+
+for i=1:nclasses
+    [rec_all(:,i),prec_all(:,i),ap_all(:,i)]=THUMOSeventclspr(test(:,i),gtlabel(:,i));
+end
+map=mean(ap_all);
+wap=sum(ap_all.*sum(gtlabel,1))/sum(gtlabel(:));
+fprintf('\n\n')
+fprintf('Per-Frame MAP: %f\n',map);
+fprintf('Per-Frame WAP: %f (weighted by size of each class)',wap);
+fprintf('\n\n')
+
+
+function [rec,prec,ap]=THUMOSeventclspr(conf,labels)
+[so,sortind]=sort(-conf);
+tp=labels(sortind)==1;
+fp=labels(sortind)~=1;
+npos=length(find(labels==1));
+
+% compute precision/recall
+fp=cumsum(fp);
+tp=cumsum(tp);
+rec=tp/npos;
+prec=tp./(fp+tp);
+
+% compute average precision
+
+ap=0;
+tmp=labels(sortind)==1;
+for i=1:length(conf)
+    if tmp(i)==1
+        ap=ap+prec(i);
+    end
+end
+ap=ap/npos;
+
+
+function [gtids,gtclasses] = load_charades_localized(gtpath,frames_per_video)
+% Loads the ground truth annotations from the csv file
+f = fopen(gtpath);
+
+% read column headers
+headerline = textscan(f,'%s',1);
+headerline = regexp(headerline{1}{1},',','split');
+ncols = length(headerline);
+headers = struct();
+for i=1:ncols
+    headers = setfield(headers,headerline{i},i);
+end
+
+% read data
+gtcsv = textscan(f,repmat('%q ',[1 ncols]),'Delimiter',',');
+fclose(f);
+ntest = size(gtcsv{1},1);
+framechar = char(cellfun(@(x) sprintf('%03d',x),num2cell((1:50)'),'UniformOutput',false')); %for speed
+gtids = cell(frames_per_video*ntest,1);
+gtclasses = cell(frames_per_video*ntest,1);
+uncell = @(x) x{1};
+c = 1;
+for i=1:ntest
+    id = gtcsv{headers.id}{i};
+    classes = gtcsv{headers.actions}{i};
+    time = str2double(gtcsv{headers.length}{i});
+    if strcmp(classes,'') 
+        missing = true;
+    else
+        missing = false;
+        classes = regexp(classes,';','split')';
+        classes = cellfun(@(x) uncell(textscan(x,'c%f %f %f','CollectOutput',true)), classes,'UniformOutput',false); %for speed
+        classes = cell2mat(classes);
+        % classes is C by 3 matrix where each row is [class start end] for an action
+    end
+    for j=1:frames_per_video
+        frameclasses = zeros(50,1); %for speed
+        fc = 1;
+        timepoint = (j-1)/frames_per_video*time;
+        for k=1:size(classes,1)
+            if missing; continue; end
+            if (classes(k,2) <= timepoint) && (timepoint <= classes(k,3))
+                frameclasses(fc) = classes(k,1);
+                fc = fc+1;
+            end
+        end
+        frameid = [id,'-',framechar(j,:)]; %for speed
+        gtids{c} = frameid;
+        gtclasses{c} = frameclasses(1:(fc-1));
+        c = c+1;
+    end
+end
+gtids = gtids(1:c-1);
+gtclasses = gtclasses(1:c-1);
+
+

Charades/Charades_v1_mapping.txt

@@ -0,0 +1,157 @@
+c000 o009 v008
+c001 o009 v016
+c002 o009 v023
+c003 o009 v025
+c004 o009 v026
+c005 o009 v030
+c006 o012 v001
+c007 o012 v006
+c008 o012 v012
+c009 o033 v016
+c010 o033 v018
+c011 o033 v018
+c012 o033 v026
+c013 o033 v030
+c014 o033 v032
+c015 o025 v008
+c016 o025 v014
+c017 o025 v016
+c018 o025 v023
+c019 o025 v024
+c020 o001 v008
+c021 o001 v012
+c022 o001 v016
+c023 o001 v023
+c024 o001 v025
+c025 o004 v001
+c026 o004 v008
+c027 o004 v012
+c028 o004 v016
+c029 o004 v019
+c030 o004 v023
+c031 o004 v025
+c032 o004 v031
+c033 o035 v008
+c034 o035 v016
+c035 o035 v023
+c036 o035 v025
+c037 o035 v026
+c038 o035 v030
+c039 o005 v001
+c040 o005 v008
+c041 o005 v012
+c042 o005 v016
+c043 o005 v023
+c044 o005 v023
+c045 o005 v025
+c046 o020 v001
+c047 o020 v008
+c048 o020 v012
+c049 o020 v016
+c050 o020 v023
+c051 o020 v031
+c052 o020 v014
+c053 o031 v008
+c054 o031 v016
+c055 o031 v003
+c056 o031 v023
+c057 o031 v028
+c058 o031 v025
+c059 o007 v018
+c060 o007 v022
+c061 o016 v008
+c062 o016 v016
+c063 o016 v023
+c064 o016 v025
+c065 o029 v005
+c066 o029 v011
+c067 o029 v008
+c068 o029 v016
+c069 o029 v023
+c070 o003 v008
+c071 o003 v016
+c072 o003 v021
+c073 o003 v023
+c074 o003 v025
+c075 o003 v026
+c076 o027 v008
+c077 o027 v016
+c078 o027 v021
+c079 o027 v023
+c080 o027 v025
+c081 o030 v016
+c082 o030 v026
+c083 o026 v023
+c084 o026 v008
+c085 o026 v009
+c086 o026 v016
+c087 o025 v013
+c088 o026 v031
+c089 o037 v001
+c090 o037 v012
+c091 o037 v030
+c092 o037 v031
+c093 o023 v008
+c094 o023 v019
+c095 o023 v030
+c096 o023 v031
+c097 o014 v029
+c098 o006 v008
+c099 o006 v016
+c100 o006 v023
+c101 o006 v025
+c102 o006 v026
+c103 o021 v006
+c104 o021 v027
+c105 o021 v027
+c106 o010 v004
+c107 o010 v008
+c108 o010 v015
+c109 o010 v016
+c110 o010 v023
+c111 o010 v030
+c112 o008 v001
+c113 o008 v012
+c114 o008 v026
+c115 o024 v008
+c116 o024 v016
+c117 o024 v023
+c118 o011 v008
+c119 o011 v016
+c120 o011 v023
+c121 o011 v030
+c122 o032 v010
+c123 o032 v018
+c124 o015 v010
+c125 o015 v018
+c126 o015 v025
+c127 o015 v026
+c128 o022 v008
+c129 o022 v005
+c130 o017 v016
+c131 o034 v009
+c132 o034 v031
+c133 o002 v000
+c134 o002 v010
+c135 o002 v018
+c136 o036 v006
+c137 o036 v008
+c138 o036 v023
+c139 o019 v030
+c140 o013 v006
+c141 o013 v007
+c142 o028 v001
+c143 o028 v012
+c144 o018 v006
+c145 o024 v032
+c146 o000 v000
+c147 o016 v002
+c148 o009 v003
+c149 o000 v009
+c150 o000 v017
+c151 o000 v018
+c152 o000 v019
+c153 o000 v020
+c154 o000 v022
+c155 o009 v028
+c156 o016 v005

Charades/Charades_v1_objectclasses.txt

@@ -0,0 +1,38 @@
+o000 None
+o001 bag
+o002 bed
+o003 blanket
+o004 book
+o005 box
+o006 broom
+o007 chair
+o008 closet/cabinet
+o009 clothes
+o010 cup/glass/bottle
+o011 dish
+o012 door
+o013 doorknob
+o014 doorway
+o015 floor
+o016 food
+o017 groceries
+o018 hair
+o019 hands
+o020 laptop
+o021 light
+o022 medicine
+o023 mirror
+o024 paper/notebook
+o025 phone/camera
+o026 picture
+o027 pillow
+o028 refrigerator
+o029 sandwich
+o030 shelf
+o031 shoe
+o032 sofa/couch
+o033 table
+o034 television
+o035 towel
+o036 vacuum
+o037 window

Charades/Charades_v1_verbclasses.txt

@@ -0,0 +1,33 @@
+v000 awaken
+v001 close
+v002 cook
+v003 dress
+v004 drink
+v005 eat
+v006 fix
+v007 grasp
+v008 hold
+v009 laugh
+v010 lie
+v011 make
+v012 open
+v013 photograph
+v014 play
+v015 pour
+v016 put
+v017 run
+v018 sit
+v019 smile
+v020 sneeze
+v021 snuggle
+v022 stand
+v023 take
+v024 talk
+v025 throw
+v026 tidy
+v027 turn
+v028 undress
+v029 walk
+v030 wash
+v031 watch
+v032 work

Charades/README.txt

@@ -0,0 +1,229 @@
+###########################################################
+     ________  _____    ____  ___    ____  ___________
+    / ____/ / / /   |  / __ \/   |  / __ \/ ____/ ___/
+   / /   / /_/ / /| | / /_/ / /| | / / / / __/  \__ \ 
+  / /___/ __  / ___ |/ _, _/ ___ |/ /_/ / /___ ___/ / 
+  \____/_/ /_/_/  |_/_/ |_/_/  |_/_____/_____//____/  
+                                                    
+###########################################################
+
+The Charades Dataset
+allenai.org/plato/charades/
+Initial Release, June 2016
+
+Gunnar A. Sigurdsson
+Gul Varol
+Xiaolong Wang
+Ivan Laptev
+Ali Farhadi
+Abhinav Gupta
+
+If this work helps your research, please cite:
+@article{sigurdsson2016hollywood,
+author = {Gunnar A. Sigurdsson and G{\"u}l Varol and Xiaolong Wang and Ivan Laptev and Ali Farhadi and Abhinav Gupta},
+title = {Hollywood in Homes: Crowdsourcing Data Collection for Activity Understanding},
+journal = {ArXiv e-prints},
+eprint = {1604.01753}, 
+year = {2016},
+url = {http://arxiv.org/abs/1604.01753},
+}
+
+Relevant files:
+README.txt (this file)
+license.txt (the license file, this must be included)
+Charades.zip:
+  Charades_v1_train.csv (the training annotations)
+  Charades_v1_test.csv (the testing annotations)
+  Charades_v1_classes.txt (the classes)
+  Charades_v1_objectclasses.txt (the primary object classes)
+  Charades_v1_verbclasses.txt (the primary verb classes)
+  Charades_v1_mapping.txt (mapping from activity to object and verb)
+  Charades_v1_classify.m (evaluation code for video-level classification)
+  Charades_v1_localize.m (evaluation code for temporal action detection)
+  test_submission_classify.txt (example test output to evaluate an algorithm)
+  test_submission_localize.txt (example test output to evaluate an algorithm)
+Charades_v1.zip (the videos)
+Charades_caption.zip (contains evaluation code for caption generation)
+Charades_v1_rgb.tar (the videos stored as jpg frames at 24 fps)
+Charades_v1_flow.tar (the flow stored as jpg frames at 24 fps)
+Charades_v1_features_rgb.tar.gz (fc7 features from the RGB stream of a Two-Stream network)
+Charades_v1_features_flow.tar.gz (fc7 features from the Flow stream of a Two-Stream network)
+Please refer to the website to download any missing files.
+
+
+###########################################################
+Charades_v1.zip 
+###########################################################
+The zipfile contains videos encoded in H.264/MPEG-4 AVC (mp4) using ffmpeg:
+ffmpeg -i input.ext -vcodec libx264 -crf 23 -c:a aac -strict -2 -pix_fmt yuv420p output.mp4
+The videos, originally in various formats, maintain their original resolutions and framerates.
+
+
+###########################################################
+Charades_v1_classes.txt 
+###########################################################
+Contains each class label (starting at c000) followed by a human-readable description of the action, such as "c008 Opening a door"
+
+
+###########################################################
+Charades_v1_train.csv and Charades_v1_test.csv
+###########################################################
+A comma-seperated csv, where a field may be enclosed by double quotation marks (") in case it contains a comma. If a field has multiple values, such as multiple actions, those are seperated by semicolon (;). The file contains the following fields:
+
+- id:
+Unique identifier for each video.
+- subject:
+Unique identifier for each subject in the dataset
+- scene:
+One of 15 indoor scenes in the dataset, such as Kitchen
+- quality:
+The quality of the video judged by an annotator (7-point scale, 7=high quality)
+- relevance: 
+The relevance of the video to the script judged by an annotated (7-point scale, 7=very relevant)
+- verified:
+'Yes' if an annotator successfully verified that the video matches the script, else 'No'
+- script:
+The human-generated script used to generate the video
+- descriptions:
+Semicolon-separated list of descriptions by annotators watching the video
+- actions:  
+Semicolon-separated list of "class start end" triplets for each actions in the video, such as c092 11.90 21.20;c147 0.00 12.60
+- length:
+The length of the video in seconds
+
+This can be loaded into MATLAB as follows:
+
+f = fopen('Charades_v1_train.csv');
+header = textscan(f,repmat('%s ',[1 10]),1,'Delimiter',',');
+csv = textscan(f,repmat('%q ',[1 10]),'Delimiter',',');
+actions = csv{10};
+actions_in_first_video = regexp(actions{1},';','split');
+
+
+This can be loaded into python as:
+
+import csv
+with open('Charades_v1_train.csv') as f:
+    reader = csv.DictReader(f)
+    for row in reader:
+        actions = row['actions'].split(';')
+
+Please refer to the evaluation code for usage examples.
+
+
+###########################################################
+Charades_v1_classify.m
+###########################################################
+Evaluation code for video-level classification. Each video has zero or more actions. This script takes in a "submission file" which is a csv file of the form:
+
+id vector
+
+where 'id' is a video id for a given video, and 'vector' is a whitespace delimited list of 157 floating point numbers representing the scores for each action in a video. An example submission file is provided in test_submission_classify.txt
+
+The evaluation script calculates the mean average precision (mAP) for the videos. That is, the average of the average precision (AP) for a single activity in all the videos.
+
+
+###########################################################
+Charades_v1_localize.m
+###########################################################
+Evaluation code for frame-level classification (localization). Each frame in a video has zero or more actions. This script takes in a "submission file" which is a csv file of the form:
+
+id framenumber vector
+
+where 'id' is a video id for a given video, 'framenumber' is the number of frame described below, and 'vector' is a whitespace delimited list of 157 floating point numbers representing the scores of each action in a frame. An example submission file is provided in test_submission_localize.txt (download this file with get_test_submission_localize.sh).
+
+To avoid extremely large submission files, the evaluation script evaluates mAP on 25 equally spaced frames throughout each video. The frames are chosen as follows
+
+for j=1:frames_per_video
+    timepoint(j) = (j-1)*time/frames_per_video;
+
+That is: 0, time/25, 2*time/25, ..., 24*time/25.
+
+The baseline performance was generated by calculating the action scores at 75 equally spaced frames in the video (our batchsize) and picking every third prediction.
+
+For more information about localization, please refer to the following publication:
+@article{sigurdsson2016asynchronous,
+author = {Gunnar A. Sigurdsson and Santosh Divvala and Ali Farhadi and Abhinav Gupta},
+title = {Asynchronous Temporal Fields for Action Recognition},
+journal={arXiv preprint arXiv:1612.06371},
+year={2016},
+pdf = {http://arxiv.org/pdf/1612.06371.pdf},
+code = {https://github.com/gsig/temporal-fields},
+}
+
+
+###########################################################
+Charades_v1_rgb.tar
+###########################################################
+These frames were extracted at 24fps using the following ffmpeg call for each video in the dataset:
+
+line=pathToVideo
+MAXW=320
+MAXH=320
+filename=$(basename $line)
+ffmpeg -i "$line" -qscale:v 3 -filter:v "scale='if(gt(a,$MAXW/$MAXH),$MAXW,-1)':'if(gt(a,$MAXW/$MAXH),-1,$MAXH)',fps=fps=24" "/somepath/${filename%.*}/${filename%.*}_%0d.jpg";
+
+The files are stored as Charades_v1_rgb/id/id-000000.jpg where id is the video id and 000000 is the number of the frame at 24fps.
+
+
+###########################################################
+Charades_v1_flow.tar
+###########################################################
+The flow was calculated similarly at 24fps using the OpenCV "Dual TV L1" Optical Flow Algorithm (OpticalFlowDual_TVL1_GPU)
+The flow for each frame is stored as id-000000x.jpg and id-000000y.jpg for the x and y components of the flow respectively. 
+The flow is mapped to the range {0,1,...,255} with the following formula:
+y = 255*(x-L)/(H-L)
+y = max(0,y)
+y = min(255,y)
+where L=-20 and H=20, the lower and high bounds of the optical flow.
+The files are stored as Charades_v1_flow/id/id-000000x.jpg where id is the video id, 000000 is the number of the frame at 24fps, and x is either x or y depending on the optical flow direction (optical flow is stored as two seperate grayscale images for the two channels)
+
+
+###########################################################
+Charades_v1_features_{rgb,flow}.tar.gz
+###########################################################
+Using the two-stream code available at github.com/gsig/charades-algorithms we extracted fc7 features (after ReLU) from VGG-16 rgb and optical flow streams using the provided models (twostream_rgb.t7 and twostream_flow.t7). Logistic regression (linear layer, softmax, and cross entropy loss) on top of these features gives 18.9% accuracy on Charades classification. Simplified code for extracting the features is as follows:
+
+fc7 = model.modules[37].output
+for i=1,fc7:size(1) do
+   out.write(string.format('%.6g',fc7[i]))
+
+There are two folders Charades_v1_features_rgb/ and Charades_v1_features_flow/ for the two streams.
+The features are stored in a whitespace delimited textfile for the 4096 numbers. 
+
+The files are stored as Charades_v1_features/rgb/id/id-000000.txt where id is the video is and 000000 is the number of the frame at 24fps, which matches the provided rgb and flow data. To limit the file size, we include every 4th frame, but the frame numbers correspond to 24fps, so the numbers are 1,5,9,13, etc.
+The features can be loaded as follows:
+
+Python: 
+fc7 = numpy.loadtxt('id/id-000000.txt')
+
+Torch: 
+file = io.open('id/id-000000.txt')
+xx = torch.Tensor(file:lines()():split(' '));
+file:close()
+
+
+###########################################################
+Baseline algorithms on Charades 
+###########################################################
+Code for multiple activity recognition algorithms are provided at:
+https://github.com/gsig/charades-algorithms
+
+
+###########################################################
+CHANGELOG
+###########################################################
+6/1/16
+Initial release
+
+2/27/17
+Adding support for evaluating localization. New evaluation script for localization. 
+'length' column was added to Charades_v1_train.csv and Charades_v1_test.csv to have an official length of each video.
+Adding details about provided RGB and Flow data.
+
+5/14/17
+Adding details about provided fc7 features.
+Improving scene annotations ('scene' column in _train and _test).
+
+
+###########################################################

Charades/get_test_submission_localize.sh

@@ -0,0 +1,4 @@
+#!/bin/bash 
+# Omitted from this zip file due to size. 49M compressed.
+wget http://ai2-website.s3.amazonaws.com/data/test_submission_localize.zip
+unzip test_submission_localize.zip

Charades/license.txt

@@ -0,0 +1,15 @@
+License for Non-Commercial Use
+
+If this software is redistributed, this license must be included. The term software includes any source files, documentation, executables, models, and data.
+
+This software and data is available for general use by academic or non-profit, or government-sponsored researchers.  It may also be used for evaluation purposes elsewhere.  This license does not grant the right to use this software or any derivation of it in a for-profit enterprise.  For commercial use, please contact The Allen Institute for Artificial Intelligence.  
+
+This license does not grant the right to modify and publicly release the data in any form.
+
+This license does not grant the right to distribute the data to a third party in any form.
+
+The subjects in this data should be treated with respect and dignity. This license only grants the right to publish short segments or still images in an academic publication where necessary to present examples, experimental results, or observations.
+
+This software comes with no warranty or guarantee of any kind.  By using this software, the user accepts full liability.
+
+The Allen Institute for Artificial Intelligence (C) 2016.

README.md

@@ -0,0 +1,61 @@
+
+# Description
+
+## Dataset V1-2
+
+1. v1-2\_train.tar.gz and v1-2\_val.tar.gz
+
+Data (train and val set) associated with ActivityNet release 1.2
+
+2. v1-2\_test.tar.gz
+
+Data (test set only) associated with ActivityNet release 1.2
+
+## Dataset V1-3
+
+1. v1-3\_train\_val.tar.gz
+
+  - *Additional* videos (train val set) collected for ActivityNet release 1.3
+
+  - v1-3 is an extension of v1-2, so you also need to download v1-2 data and merge to v1.3 
+
+2. v1-3\_test.tar.gz
+
+  - *Additional* videos (test set only) collected for ActivityNet release 1.3
+
+  - v1-3 is an extension of v1-2, so you also need to download v1-2 data and merge to v1.3 
+
+## Missing files
+
+For people who download the dataset from youtube-dl, we provide the missing videos.
+
+1. missing_files_trainval.zip
+
+  - Missing videos in trainval set
+
+2. missing_files_v1-2_test.zip
+
+  - Missing videos in v1-2 test set
+
+3. missing_files_v1-2_test.zip
+
+  - Missing videos in v1-2 test set
+
+## H5 sum
+b06ca628443da1c1389c69b2feac6ca2  v1-2_test.tar.gz
+8510ed0d8ae399b1bda399fd117e7406  v1-2_train.tar.gz
+34f0d2b867f188883259b607e68c830f  v1-2_val.tar.gz
+ec218c66991d776ad0ff826c0b43836d  v1-3_test.tar.gz
+eb0829a2eae2e3f254a0fcb64c6583c4  v1-3_train_val.tar.gz
+
+## External Link
+
+For people who are not able to download large files from google drive, we give a link to another driver. 
+
+1. Baidu Pan (please send me an email if it is expired)
+
+    - link: https://pan.baidu.com/s/1Ic6wrvOUJARgnqRSk-BzjA 
+
+    - code: f766
+
+

missing_files.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0dc600595a22645b37e203cb2bf6d65f733c91614c67b2846a4b414b5709785
+size 37798739657

missing_files_v1-2_test.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35982454ddbc4a08f1ef65c20e25c0d4afbd6ce8d3a6ad041668bec0a1f57432
+size 5687952619

missing_files_v1-3_test.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f82e05f37ca28562d4c5acb8d5e7cf8c25c5696efc1d978616ca07ab4328f6e
+size 6888567987

v1-2_test.tar.gz.00

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f361c3f201745f1951b7f6dd22069085db04736c3d19eeb4c72ac35ebd9cf482
+size 4194304000

v1-2_test.tar.gz.01

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6577c76853b0e25a8cc0144ee867c62f4c9f229be76458b15e7f242a1e6f78f2
+size 4194304000

v1-2_test.tar.gz.02

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf7fef61c7c8e4d202cfa5702b0ade27fa203685598501337da00bba62725723
+size 4194304000

v1-2_test.tar.gz.03

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c839e73be8fe894b8e21356a62cb3fcba8fbf0766bd88b3ce953d0c851ec269
+size 4194304000

v1-2_test.tar.gz.04

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4e76629544143b504593b993ea67cb5a014fa391e0ad3d5e455342c6fc1c767
+size 4194304000

v1-2_test.tar.gz.05

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45840394be695f15478c47e682428598a540a33e800ad5903bd84c3a5dda8692
+size 4194304000

v1-2_test.tar.gz.06

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1170e038959c78333ba85b309ae8b8bd0881296f51949d61342b880991d880bc
+size 4194304000

v1-2_test.tar.gz.07

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ffb1751ec8bd443faebabedf17a7e5312d59e58200fd8b53b18bc81072126fd
+size 4194304000

v1-2_test.tar.gz.08

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f04b66acab5f8451659c9db4bd322601ba54cbda5012b3eaabd74f6ee2807208
+size 4194304000

v1-2_test.tar.gz.09

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a52e93082da25b6bec812e26e99085e326004e7278bc2d880313806c86d72ed
+size 4194304000

v1-2_test.tar.gz.10

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:59c396b070db0df6af456088f0b83081b1d60c72cc71d49e86cc2100850852d8
+size 4194304000

v1-2_test.tar.gz.11

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9a829bea0e3cf26e83e9e80c4d1d7a5cf07576af91c695188f88aa7fadffbea
+size 4194304000

v1-2_test.tar.gz.12

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c49eeec20519d95a0956cbe0f85d15c96e4ce3898e0576ff04bb276d334e0b7e
+size 4194304000

v1-2_test.tar.gz.13

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4155fcf720960e4f35769d1a117382e3b9f05df0d9ad583637a46f3c140a082e
+size 3543497960

v1-2_train.tar.gz.00

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14c7d31d3d34e6d2c27240a69d6734528295bd497f7d6aed67f6f7f3a4ac83ab
+size 4194304000

v1-2_train.tar.gz.01

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6a3a6bc2369b450b08e911306e0780f6dad11d8090c3143f26dd4453ee50ee9
+size 4194304000

v1-2_train.tar.gz.02

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4cf214623928c4dae110887bcc978a210de96777f7c14a8a661e5e4c9ae417f
+size 4194304000

v1-2_train.tar.gz.03

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89d3b7b6ad10bda6894b001768763de1e08ee69e28b08de0795d93b7a6107643
+size 4194304000

v1-2_train.tar.gz.04

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:876e9c01fe9dc6d5bf868f2e3760af2eed9e08a2dd08eb90230fab79657460ae
+size 4194304000

v1-2_train.tar.gz.05

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edee3fff13ea5ecba52c7ed9783ceff179dc8598de07cd23b18c6fabd1d3fb66
+size 4194304000

v1-2_train.tar.gz.06

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8602441bd94107267cb97ffaa2be62be474edda59c2ec1e638a0867fd4dbe796
+size 4194304000

v1-2_train.tar.gz.07

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a89792fb4aea0fbeade48de320c6e36bb8c352b87766f4996730605823725be2
+size 4194304000

v1-2_train.tar.gz.08

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a52ac47c177ca2668a1f05ef06393ec2445ad743e5200945df1d9bbe6bc13ed9
+size 4194304000

v1-2_train.tar.gz.09

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2a91408eed8a206b5660aaec9ede6fbbecd1fc94a434edb5537403f03def26d
+size 4194304000

v1-2_train.tar.gz.10

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7935b6b56fbcfcc7fb74cbabb6e859c1c11d01c2581d2f2ad52734fd039953c5
+size 4194304000

v1-2_train.tar.gz.11

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bb85c04f2e7ad06605ce6c7e1d56fb61f4860386994870c91c99851c6c62f83
+size 4194304000

v1-2_train.tar.gz.12

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dc08886befd6e09d800b9ca0eac973d79c03ec87908425a0af9d03a62faa9ae
+size 4194304000

v1-2_train.tar.gz.13

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7bcec5818adca6022245a89ba55218ec9ee23d2c0fd77718b38d7258bc30e43
+size 4194304000

v1-2_train.tar.gz.14

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09cf2292e6e4899da1ebcec28155f196364b501cd503dfd22ada285369994d44
+size 4194304000

v1-2_train.tar.gz.15

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cadf8631e09ec1c6bd1f5629534d926420a504151b1760dee46767ddc9d1479
+size 4194304000