API Reference

Organisation

Get data row

GET https://app.picsellia.com/api/V1/dataset/:dataset_id/:row_id

Retrieve all the informations about an asset from one of your dataset

Path Parameters

Name	Type	Description
dataset_id	string	The Id of the dataset you want to retrieve data from

Headers

Name	Type	Description
API Token	string

Request Body

Name	Type	Description
raw_id	string	id of the data
filename	string	filename of the data

200

{
    'id': '8e6487ae-6f23-4d62-8bdd-8c808894fbe1',
    'external_url': 'my_awesome_picture.jpg',
    'object_name': 'b5620ac0-cee8-4fd8-b81a-308657558848%5CPHOTO%20843.jpg'
    'type': 'image',
    'format': 'jpg',
    'size': 3.4,
    'import_date': '05-28-2020',
}

Projects

Create project

POST https://app.picsellia.com/api/v1/create_project

Allows you to setup a whole project at once

Headers

Name	Type	Description
API Token	string	Your personnal API token available in your profile settings/API token

Request Body

Name	Type	Description
type	string	'classification' or 'detection' or 'segmentation' depending on the task
team	object	Dictionnary with the emails of accounts you want to add on your project and the desired roles. If some of the emails doesn't exists, people will receive an invitation to join the platform and the project. { '1': { 'email': '[e-mail]', 'role': 'labeler' or 'reviewer' or 'manager' }, '2': { ... }, ... }
description	string	The description of your project
private	boolean	Set it to True if you want your project to be private
labels	object	{ '1': { 'label_name': 'label', 'tool': 'rectangle' or 'polygon' or 'line' or 'point' }, '2': { ... }, ... }
dataset_name	string	Name of the dataset you want to attach to your project (must be already created)
project_name	string	Must be different from your other project's names

200

Get project list

GET https://app.picsellia.com/api/v1/project_list

Retrieve the list of your project names and ids

Headers

Name	Type	Description
API Token	string	Your personnal API Token

200

{
    '1': {
        'project_name': 'my_first_project',
        'id': 'e403f741-3672-4a7a-a39d-34e8be241dd2'
    },
    '2': {
        ...
    }
}    

Get project infos

GET https://app.picsellia.com/api/v1/project_infos/:project_id

Retrieve all the informations about one of your project

Path Parameters

Name	Type	Description
project_id	string	The id of the project you want to retrieve informations from

Headers

Name	Type	Description
API Token	string	Your personnal API Token

200

{
    'id': 'e403f741-3672-4a7a-a39d-34e8be241dd2',
    'project_name': 'my_first_project',
    'type': segmentation,
    'dataset': {
        'dataset_id': 'a0c2b65f-6e25-445d-a423-22b7ff69f664',
        'dataset_name': 'awesome_planes'
    },
    'labels': {
        '1': {
            'label_name':  'plane',
            'tool': 'polygon'
        },
        '2': {
            'label_name':  'bird',
            'tool': 'polygon'
        }
    },
    'collaborators': {
        '1': {
            'email': '[e-mail]',
            'role': 'admin'
        },
    },
    'models': {
        '1': {
            'model_id': '3c36ba70-faa7-41bf-b17e-d2727d30edc0',
            'model_name': 'MaskR-CNN_plane_detection'
        },
    }
    'metrics': {
        'nb_annotations': 13200,
    }
}

Update project infos

PUT https://app.picsellia.com/api/v1/project_infos/:project_id

Update the informations (project name, description) of one of your projects

Path Parameters

Name	Type	Description
project_id	string	The ID of the project you want to update

Headers

Name	Type	Description
API Token	string	Your personnal API Token

Request Body

Name	Type	Description
description	string	The description of your project
project_name	string	The project_name you want for your project

200

Get project collaborators

GET https://app.picsellia.com/api/v1/project_collaborators/:project_id

Retrieve the list of collaborators for a given project

Headers

Name	Type	Description
API Token	string

200

{
    '1': {
        'email': '[e-mail]',
        'username': 'the_john',
        'role': 'manager',
        'date_joigned': '12-06-2020'
    },
    '2': {
        'email': '[e-mail]',
        'username': 'gina',
        'role': 'reviewer',
        'date_joigned': '14-06-2020'
    },
}

Update project collaborators

PUT https://app.picsellia.com/api/v1/project_collaborators/:project_id

Update the list of collaborators on a given project

Path Parameters

Name	Type	Description
project_id	string	The Id of your project

Headers

Name	Type	Description
API Token	string

Request Body

Name	Type	Description
collaborators	object	An object containing the updated list of collaborators for this project. { '1': { 'email': '[e-mail]', 'role': 'admin' }, '2': { 'email': '[e-mail]', 'role': 'reviewer' }, '3': { 'email': '[e-mail]', 'role': 'manager' }, }

200

Update project labels

PUT https://app.picsellia.com/api/v1/project_labels/:project_id

Update the label list of a given project

Path Parameters

Name	Type	Description
project_id	string	The Id of your project

Headers

Name	Type	Description
API Token	string

Request Body

Name	Type	Description
labels	object	An object containing the updated list of your project's labels. { '1': { 'label_name': 'plane', 'type': 'polygon' }, '1': { 'label_name': 'bird', 'type': 'polygon' }, '3': { 'label_name': 'wind-turbine', 'type': 'polygon' }, }

200

Set project automations

PUT https://app.picsellia.com/api/v1/project_auto/:project_id

Attach the labels of your project to the labelmap of a chosen pre-trained network to enable the 'pre-annotation' in our annotation interface (The default network is Mask R-CNN trained on COCO).

Path Parameters

Name	Type	Description
project_id	string	The Id of your project

Headers

Name	Type	Description
API Token	string

Request Body

Name	Type	Description
model_id	string	The Id of the pre-trained model you want to associate your labels to.
labels	object	An object containing the list of your project's labels and the associated pre-annotation label. { '1': { 'label_name': 'plane', 'link_with': 'plane' }, '2': { 'label_name': 'bird', 'link_with': 'bird' } }

200

The model that you want to associate to your labels must be already attached to your project !

See Attach a model to my project

Set project Questions & Answers

PUT https://app.picsellia.com/api/v1/project_qa/:project_id

Allows you to setup the Questions & Answers feature for your project. There is 4 different type of answers available : - Text answer : The user will answer the question via a text input - Value selection answer : The user will answer the question by choosing a value in a selection menu - Multiple Choice answer : The user will answer the question by checking some checkboxes - Range answer : The user will answer the question by moving a cursor in the desired range You can setup an image-level Q&A, which means that the questions are asked relatively to the image, and you can also set shape-level Q&A where every question will be asked for each drawn shapes.

Path Parameters

Name	Type	Description
project_id	string	The Id of your project

Headers

Name	Type	Description
API Token	string

Request Body

Name	Type	Description
shape_qa	object	Same as image_qa but will be asked for every shapes
image_qa	object	The image level Q&A { '1': { 'question': 'What is the person doing ?', 'type': 'text' }, '2': { 'question': 'How many people are in the image ?', 'type': 'range', 'min': 0, 'max': 50 } '3': { 'question': 'Level of brightness ?', 'type': 'choice', 'values': [0,1,2,3,4] } '4': { 'question': 'What emotions is the most probable?', 'type': 'mc', 'values': ['sadness','anger','happiness','surprise','joy'] } }

200

Get project stats

GET https://app.picsellia.com/api/v1/project_stats/:project_id

Retrieve the current state and statistics of a given project

Path Parameters

Name	Type	Description
project_id	string	The Id of your project

Headers

Name	Type	Description
API Token	string

200

{
    "labels": ["motor", "bus", "awning-tricycle", "tricycle", "truck", "van", "car", "bicycle", "people", "pedestrian", "others"], 
    "label_variance": 43.3, 
    "predicted_time": 0, 
    "time_per_image": [3, 17, 11, 14, 21, 3, 24, 3, 1, 15, 6, 22, 7, 16, 3, 7, 5, 23], 
    "assets_per_image": [127.0, 77.0, 48.0, 115.0, 113.0, 140.0, 115.0, 136.0, 164.0, 31.0], 
    "images_remaining": 0, 
    "total_time_spent": "", 
    "labels_per_classe": [1637.0, 20.0, 204.0, 266.0, 111.0, 286.0, 2144.0, 214.0, 1830.0, 2329.0, 6.0], 
    "nb_images_annotated": 150, 
    "average_time_by_image": 239.6, 
    "mean_assets_per_images": 60.31333333333333, 
    "percentage_advancement": 100, 
    "total_labels_submitted": 9047
}

Attach model to project

POST https://app.picsellia.com/api/v1/attach_model/project_id/model_id

Attach a CNN model to your project to perform transfert learning of

Path Parameters

Name	Type	Description
model_id	string	id of your model
project_id	string	id of your project

200

"Model attached"

202

Delete project

DELETE https://app.picsellia.com/api/v1/delete_project/project_id

Delete a project

Path Parameters

Name	Type	Description
project_id	string	id of the project

200

"Project deleted"

401

Detach model from project

PUT https://app.picsellia.com/api/v1/delete_project/project_id/model_id

Detach a model from a project

Path Parameters

Name	Type	Description
model_id	string
project_id	string	id of the project

200

"Model detached"

Check connection

GET https://app.picsellia.com/sdk/check_connection

This endpoint allows you to check connection with our Backend and assert your Token is valid.

201 Connection successfully instantiated

{"project_id": (uuid) id of the project}

400 Could not find matching project for this token

GET

Path Parameters

Name	Type	Description
	string

200

Models

Get model list

GET https://app.picsellia.com/api/v1/models_list

Retrieve the list of all the models attached to a project or created where you are manager level

Request Body

Name	Type	Description
	string	Name of your model or desired name for model creation

200

{
        '0': {
            'model_id' : id
            'model_name':  'Awesome model',
            'model_type': 'classification', 'segmentation', 'detection',
            'tags' : ["...", "...", "coco"]
            'used_in': {
                  '0': {
                  'project_id': id,
                  'project_name': 'Awesome model',
                  'project_type': 'classification', 'segmentation', 'detection',
                  },
                  '1': {
                  ...
                  }
            },
            'predictable' : True or False,
            'trainable' : True or False,
            },
            
        '1': {
            ...
        }
    
    
}

202

Delete a Model

DELETE https://app.picsellia.com/api/v1/delete_model/model_id

Path Parameters

Name	Type	Description
model_id	string	id of the model

200

"Model deleted"

401

Get model infos

GET https://app.picsellia.com/api/v1/model_infos/model_id

Path Parameters

Name	Type	Description
model_id	string	id of the model

200

{'model_id' : id,
 'model_name': name,
 'model_type': type,
 'tags': [ "..", "...", ],
 'description': '...',
 'private' : True or False,
 'trainable': True of False,
 'predicatble': True of False,
 'paper_link' :'',
 'github_link': '',
 'performances': {
 
             ...
                 },
 'labels' : {
 ...
 }
 'used_in': { 
             '0' : {
               'project_id' : id,
               'project_name': name,
               },
             '1' : {
             
             ...
             
             },
             },
 }'
           
             

Update model infos

PUT https://app.picsellia.com/api/v1/model_infos/model_id

Path Parameters

Name	Type	Description
model_id	string	id of the model

Request Body

Name	Type	Description
	string
github_link	string	git repo of your model
paper_link	string	scientific paper link of the model
private	boolean	set True for private model, false for public
tags	array	array of tags in string format
description	string	description of the model
model_name	string	new_name of the model

200

Delete experiment

DELETE https://app.picsellia.com/api/v1/delete_experiments/project_id/model_id/training_id

Method to delete a version of an experiment

Path Parameters

Name	Type	Description
project_id	string	id of the project
training_id	integer	id of the experiments
model_id	string	id of the model

200

Annotations

A range of annotation

GET https://app.picsellia.com/api/v1/annotations/project_id

This allow you to retrieve annotations for a range of images.

Path Parameters

Name	Type	Description
project_id	string	id of the project

Request Body

Name	Type	Description
range	array	desired range ( [0, 15])

200

 { 
 '0': {
      "internal_picture_id": (uuid) id of the related image,
      "nb_labels": (int) nb of objects in the pic,
      "time_spent": (float) time spend to annotate the pic in sec,
      "QandA" :  {
      
      }
      "annotations": (list(dic))
       [
         {
           "type": (str) type of label -> rectangle/polygon
           "label": (str) label given to shape,
           "polygon": {"geometry": [{"x": (int) x, "y": (int) y}, ...,]
              # each dict in geometry is a point of the polygon
                # if no polygon have been done, list is empty
           "rectangle": {
                         "height": (int) h,
                         "width": (int) w,
                         "top": (int) x,
                         "left": (int) y,
                       }
          },
         ...,
      ]
   },
   
    '1' : {
    
     ...
     
      },
   } 

400

annotations

POST https://app.picsellia.com/sdk/upload_annotations

Request Body

Name	Type	Description
format	string	format type ("picsellia" of "other")

201

{"message": "Annotation uploaded"}

400

Training logs

POST https://app.picsellia.com/sdk/post_logs

The endpoints allow you to post the repartition of your train test ensemble to visualize it on the Platform.

Request Body

Name	Type	Description
training_id	integer	version of your experiments
network_id	string	id of your network
logs	object	(dict) logs from tfevents converted to dict

201

{"message" : "Metrics saved"}

400

Preview results

POST https://app.picsellia.com/sdk/post_preview

This endpoint allow you to post visual results of your experiments

Request Body

Name	Type	Description
training_id	integer	version of your experiments
network_id	string	id of your Model
urls	array	(str) url to your example results

200

Dataset

Create Dataset

POST https://app.picsellia.com/api/v1/create_dataset

API endpoint to create a Dataset

Request Body

Name	Type	Description
private	boolean	True for private, False for Public
description	string	Description of the Dataset
dataset_name	string	Desired name of the dataset

200

{"dataset_id": id}p

Get dataset list

GET https://app.picsellia.com/api/v1/dataset_list

Get the list of all the dataset of the project you are working on

Headers

Name	Type	Description
token	string	Auth token

200

{
   '0': {
    'dataset_id': id,
    'dataset_name': 'my awesome dataset',
    'dataset_size': 1000,
    'created_by': 'Username',
    'description': 'bla bla bla',
    'used_in': {
       '0': {
            'project_id': id,
            'project_name': 'awesome project 1',
            'project_type': 'classification', ' 
             }
    },
   },
   '1': {
      ...
   }
}

Dataset infos

GET https://app.picsellia.com/api/v1/dataset_info/dataset_id

Retrieve the information related to a dataset

Path Parameters

Name	Type	Description
dataset_id	string	id of the dataset

200

{ "dataset_id": id,
  "dataset_name": "awesome dataset",
  "dataset_size": 1000,
  "description": "bla bla bla",
  "used_in": {
    "0": {
    "project_id": id,
    "project_name": "awesome project",
    },
    "1" : {
      ...
      },
    },
}

Update dataset infos

PUT https://app.picsellia.com/api/v1/dataset_info/dataset_id

Update dataset infos

Path Parameters

Name	Type	Description
dataset_id	string	id of the dataset

Request Body

Name	Type	Description
description	string	new description
dataset_name	string	new name

200

"Info updated"

Delete Dataset

DELETE https://app.picsellia.com/api/v1/dataset/dataset_id

Path Parameters

Name	Type	Description
dataset_id	string	id of the dataset to delete

200

Retrieve info of a range of raw data

GET https://app.picsellia.com/api/v1/raw_data/dataset_id

Path Parameters

Name	Type	Description
dataset_id	string	id of the dataset

Request Body

Name	Type	Description
range	array	range of index to retrieve ([0:50]) for example

200

{
    "0" : {
    "id": id of the raw data,
    "external_url": url,
    "shape": [x,y],
          },
          
    "1" : {
       ...
    },
    
    ...
    
  }

Raw data

DELETE https://app.picsellia.com/api/v1/raw_data/dataset_id

you can delete a raw data either by providing its filename of the id of the datat

Path Parameters

Name	Type	Description
dataset_id	string	id of the data

Request Body

Name	Type	Description
filename	string	name of the picture to delete
raw_id	string	id of the raw data

200