End to end learning for autonomous driving on unpaved roads : a study towards automated wildlife patrol
The aim is to investigate the technological feasibility of deploying automated Unmanned Ground Vehicles (UGV) for automated wildlife patrol. The presentation outlines a feasibility study and cost benefit analysis for a project that would cover peak seasons when there is a shortage of park rangers. Data is drawn from field tests in Nairobi National Park; Ruma National Park; and on paved road/highways in Kenya).
Auteur ou autrice(s) : Brahmbhatt, Khushal, Ojino, Ronald
Télécharger le PDF
Langage : Anglais
|
Effective creation of ground truth data-set for Malaria diagnosis
The goal is to reduce mortality rates related to malaria, particularly in marginalised communities. Several Artificial Intelligence (AI) techniques have been used to solve challenges in the existing malaria diagnosis tools. The presentation summarizes the project, which has collected a dataset of 10,000 blood sample images (stained blood smear). Using the data, an open-source annotation tool was created. Working with the AI4D (Artificial Intelligence for Development) the final aim is to develop a mobile application that will assist lab technologists in malaria diagnosis.
Auteur ou autrice(s) : Shaka, Martha
Télécharger le PDF
Langage : Anglais
|
Computer vision tomato pest assessment and prediction tool
Tomato pest Tuta absoluta is a threat to tomato farmers across the world. This brief presentation analyzes images of tomato seedlings in terms of an Inception-v3 model.
Auteur ou autrice(s) : Pastory, Denis
Télécharger le PDF
Langage : Anglais
|
Zindi : all AI4D challenges
The presentation shows the competition overview (marketing and communication; and competition engagement) for the AI4D - African Language Dataset Challenge, an effort to incentivize the creation, organization and discovery of African language datasets through a competitive challenge. Hundreds of thousands of page views for the challenge are noted.
Auteur ou autrice(s) : Orlic, Davor
Télécharger le PDF
Langage : Anglais
|
Building a network of excellence in artificial intelligence in Sub-Saharan Africa : final technical report
The report provides a summary of project activities, milestones, achievements, outcomes and outputs. The project drew from the Global South ecosystem mapping, and facilitated a bottom-up network/community of researchers who investigated and provided recommendations on how future artificial intelligence for development (AI4D) networks’ research should be shaped. The project acknowledges the growing importance of funding Networks and domain-specific research relevant to African researchers and practitioners, with a focus on the ethical, legal and social aspects of AI research. One of the major results works towards addressing language barriers through machine translation.
Auteur ou autrice(s) : Orlic,Davor
Télécharger le PDF
Langage : Anglais
|
Early detection of preeclampsia using ambulatory blood pressure monitoring using wearable devices and long short term memory networks (LSTM-NN) on the edge
The study explores the use of wearable devices for ambulatory blood pressure data collection, for use in blood pressure prediction/forecasting using Long Short Term Memory (LSTM) recurrent neural networks (RNN) on mobile devices. A mobile app was developed (wearable as a smart watch) during this first phase. LSTM Network is an advanced RNN, a sequential network that allows information to persist, meaning that it has the capacity to handle the vanishing gradient problem faced by RNN. [See also https://www.k4all.org/wp-content/uploads/2019/07/Early-detection-of-pre…]
Auteur ou autrice(s) : Thiga, Moses M, Kimeto, Pamela
Télécharger le PDF
Langage : Anglais
|
Semi-automatic tool for meta-data extraction from Malawi court judgments
The project aims to build a software tool for identifying and extracting the general structure of criminal court judgments in Malawi. The combination of indexes, thesaurus of legal terms, an organised system of citation/ referencing is crucial to legal research. The MalawiLII digitized legal filing system is not complete and not easily searchable (most judgments are images of scanned pdfs). There is no linkage between case citations and no linkage from cases to Laws/ Statues. It is not trivial to select legal keywords for International Classification of Crime for Statistical Purposes (ICCS) classification. This presentation outlines the project activities and timeline.
Auteur ou autrice(s) : Taylor, Amelia
Télécharger le PDF
Langage : Anglais
|
Using artificial intelligence to digitize and analyze parliamentary bills in Sub-Saharan Africa
Derived from “Prediction and Analysis of Law-Making in Kenya,” out of 460 bills introduced to the Kenyan parliament from 2009 - 2019, only 65 (14.1%) were enacted. Using a combination of handcrafted and text-based features, the study developed a machine learning algorithm to predict the probability of a bill becoming law. Given the challenges and precariousness facing developing countries, the quality of policy making and legislation is enormously important. For instance, despite a high level of compensatory pay of Kenyan government representatives, most legislative duties and bills introduced to parliament are shrouded in mystery.
Auteur ou autrice(s) : Akinfaderin, Adewale
Télécharger le PDF
Langage : Anglais
|
Arabic speech to Moroccan sign language translator ‘Learning for Deaf’
The presentation depicts a framework for establishing a database speech-to-text for those with impaired hearing in Morocco. There is a lack of data resources about the Arabic sign language. The researchers dedicated much energy to collecting datasets of ~100 initial sign phrases. The dataset is composed of subtitled videos. The presentation includes some information on coding as well as video representation.
Auteur ou autrice(s) : Mahmoudi, Abdelhak
Télécharger le PDF
Langage : Anglais
|
Responsible artificial intelligence in Sub-Saharan Africa : landscape and general state of play
The paper reviews AI adoption and investment in Africa. An artificial intelligence (AI) ecosystem exists in Africa, with evidence of pockets of activity that could be prioritized, coordinated, and funded to result in an energized and productive AI infrastructure. The paper suggests a focus on four core areas: 1) innovation; 2) capacity building; 3) policy; and 4) infrastructure. It addresses critical issues that cut across these dimensions including political participation; scaling out; and inclusion, particularly as it relates to gender. Mauritius, positioned as 45th in the world, South Africa (59th), Seychelles (68th), Kenya (71st), and Rwanda (87th): these are Africa’s top five countries in the 2020 global Government AI Readiness Index.
Auteur ou autrice(s) : Gwagwa, Arthur, Kachidza, Patti, Siminyu, Kathleen, Smith, Matthew
Télécharger le PDF
Langage : Anglais
|
Artificial intelligence needs assessment survey in Africa
Key results of the survey are presented in this report. The findings aim to bridge the information gap concerning strategic priorities, policy measures, developmental challenges, human and institutional capacity needs, and legal frameworks concerning AI in African countries. This needs assessment publication is in line with UNESCO’s operational strategy for the implementation of its Global Priority Africa, in particular the flagship programme Number 3 “Harnessing STI and knowledge for the sustainable socio-economic development of Africa.” Potential benefits and risks of technology are not equally distributed across regions of the world.
Auteur ou autrice(s) : Sibal, Prateek, Naupane, Bhanu
Télécharger le PDF
Langage : Anglais
|
Artificial intelligence capacity in Sub-Saharan Africa : compendium report
Artificial Intelligence for Development (AI4D) Africa supported research on three key stakeholders involved in AI capacity building in sub-Saharan Africa (SSA): Centres of Higher Education and Training; Governments; and the broader AI community in the region. A comprehensive desktop literature review was also conducted. Taking available data into account together with the desktop review of literature and the findings from the UNESCO report, this detailed paper provides takeaways from the research. In line with the recommendations presented here, the key to creating a vibrant AI ecosystem is committed stakeholder engagement that pushes individual countries forward by building capacity in key areas.
Auteur ou autrice(s) : Butcher, Neil, Wilson-Strydom, Merridy, Baijnath, Mohini
Télécharger le PDF
Langage : Anglais
|
