With the present and future surveys for exoplanet detection by transit photometry (K2, TESS and PLATO), one of the biggest challenge of the next decade for the exoplanet community will be to maximise the scientific return from this huge amount of data. These missions will give us hundreds of thousands of light-curves, amongst which approximatively 10 % will exhibit transit signals. Without any vetting, around 90 % of them will be false positive, in the sense that they will not correspond to an exoplanet transiting the target star. So a key step will be to focus the follow-up ressources (in terms of telescope time, computer time and human ressources) one the planetary transit.In order address this problematic, we developed a software called BART which stands for Bayesian Analysis for the Ranking of Transits. When most of the existing ranking tools rely on a set of individual criteria, our tool performs a fully coherent analysis combining the most used criteria and taking into account their correlations.It is automated and relies on bayesian model comparison, bayesian parameter inference and Monte Carlo Markov Chain exploration technique. It produces a ranked list of transits where those which are the most likely to be due to planets (around the target stars) have a higher rank allowing to focus the follow-up on targets with the highest scientific merit.But it also gives the most probable scenario for the transits and provides an estimate of the parameters for each possible scenarios. This is a priceless information since it allows to estimate the precision required for a given follow-up observation. Thus it will allow to optimize the follow-up strategy and direct the observations.This tool has been originally developed and tested with CoRoT and I will present the results obtained on the CoRoT transit list. I will also show the results recently obtained with K2 data.< Réduire