Training camps Academic Year 2019-2020

Training camps Year 2019-2020

Training Camp Winners

Poste Italiane July 20-22

"For Challenge 1 on "Object Detection on satellite imagery": Team 6"

EleftheriaTetoula Tsonga 
Leandro Bernardino Gentili
Ilaria Taddei
Arya Farkhondeh 
"For Challenge 2 on "Big data for Smart City": Team 12"
Negin Amininodoushan
Lorenzo Ceccomancini
Salwa Ali Hersi Afrah
Flaminia Matteucci 
Grancesco Paparella 
Andrea Formichetti

Google September 3-5

"Teams Starter"

Gold  [DreamTeam], Onur Copur, Lev Telyatnikov, Flaminia Spasiano
Silver [GMR], Giusy Colarusso, Riccardo Ceccaroni, Mario Dhimitri
Bronze [DSD], Alexandru Melnic, Dilara Isikli, Dario Russo


Gold [The_WonderFul_LFM], Loretucci Lorenzo, Francesco Romeo, Mattero Zmyslowski
Silver [TheNeutralTrio], Beatrice Nobile, Lucia Testa and Edoardo Cantagallo 
Bronze [Oksana.U], Oksana ustenko


  • Mykhaylo AndriIuka, Google Research, September 3-5, on "Building an image search engine". Registration expires August 15th


How could we build an automated system that can find a photograph in a family album or an online photo collection given just a textual description? In this course we will cover fundamental techniques in computer vision and natural language processing that will help us to address this question. The main aim of the course will be to enable students to build their own prototype of the image search engine, and participate in online Kaggle competition organized for the course participants. To aid the students in this mission we will review common methods for representing images and text with neural networks.

Specific techniques that the students will learn:
- image representation with convolutional neural networks (CNNs)
- building recurrent neural networks with LSTM and GRU units
- generating natural language image descriptions (=image captioning)
- representing words and sentences with vector embeddings (Word2Vec, GloVe, and BERT)

The course will assume that the students have solid knowledge of Python and numerical computation package NumPy. Knowledge of neural network libraries such as TensorFlow and PyTorch is highly recommended, but not strictly required. We will provide self-study materials for students to catch-up on these libraries.



  • Poste Italiane July 20-22. Registration is now closed.


A network uniting the country The Poste Italiane Group constitutes the largest service distribution network in Italy. Its activities range from letter and parcel delivery to financial and insurance services, payment systems and mobile telecommunications. With its 158-year history, a network of more than 12,800 post offices, a workforce of approximately 126 thousand, total financial assets of €536 billion and 35 million customers, Poste Italiane is an integral part of Italy’s economic, social and productive fabric, occupying an unparalleled position in the country in terms of size, recognisability, reach and customer loyalty. In February 2018, Poste Italiane launched its new five-year Strategic Plan, Deliver 2022, which aims to maximise the value of the distribution network and take advantage of the market opportunities offered by digital transformation. This will involve a reorganisation of the mail and parcels segment, the expansion of financial services, consolidation of our leadership in the insurance sector and the development of payment systems. The Plan envisages investment of €2.8 billion, focusing on innovation in order to assist citizens, businesses and the Public Administration through the transition to the digital economy and offering increasingly innovative services, in line with the highest market standards. In 2019, the Group’s activities involving the production and delivery of goods and services generated direct, indirect and induced impacts on the Italian economy amounting to approximately €12.5 billion in terms of GDP. An estimated total of 189 thousand workers were employed along the production chain, resulting in the distribution of income amounting to €7.5 billion to workers.

Challenge 1 - Object Detection on satellite imagery

The purpose of this activity is to implement and train a predictive model that uses computer vision techniques to perform detection and classification of objects (buildings) using satellite images in urban areas. Students will be provided with a training labeled dataset for some urban areas, in order to train the model to recognize the different types of buildings in the aerial photos. Once the model is trained they will use it to implement the following two tasks:

·         Object detection Task: assigned a study area, students will have to identify the zones in which there is likely to be a resident population, identifying residential buildings on the provided images

·         Objects Classification task: always in the same study area, they will have to identify the number and type of residential buildings, in order to estimate the resident population

Challenge 2 – Big data for Smart City

Poste Italiane wants to play an active role by contributing to shaping and designing the Smart City, the place where everybody will live in the future, inspired by principles like as sustainability. 

The Poste Italiane mission is to contribute in making our country a better place, by promoting innovative solutions that leverage its own unique potential: from smart mobility and smart delivery, to the Post Office of the future; from PA services to digital payments.

We believe that data are at the hearth of this change and the ability to gain insights from data could enable new services and also a new way of living for our customers and citizens.

We want you to imagine and propose an innovative solution, or more integrated solutions, for Poste Italiane in the wide field of Smart City, based both on the big data assets already available in Poste Italiane and on the potential big data generated by further future innovative solutions.

The solution must be described together with a business case that highlights:

·         Main Assumptions and Rationale

·         Underlying technical logics

·         Analysis of the economic feasibility and the principal service costs of the solution

Required skills:

·         Wide knowledge of Python and NumPy stack (NumPy, SciPy, Pandas, Matplotlib)

·         Students should have a practical experience in using at least one of the neural network libraries such as Tensorflow, Keras and PyTorch

·         Basic knowledge of image processing libraries such as OpenCV (highly recommended) or Pillow.

·         Theoretical knowledge of modern computer vision techniques (CNN) with a particular focus on neural networks for object detection (RCNN, FCN, SSD, ResNet, YOLO)

·         Big Data analytics competence and technical competence in the definition of Big Data analytics platforms

·         Ability to define and describe Business Cases and Case Analysis in the wider context of use of Big Data

·         Ability to use Design Thinking methodology applied to the context