ExoMiner++ 2.0: Vetting TESS Full-Frame Image Transit Signals

TL;DR

This work extends ExoMiner++ to TESS FFIs, addressing the increased cadence and crowding challenges by adapting the model to FFI light curves and enriching inputs with neighboring-star information and high-resolution difference images. By employing multi-source learning that combines 2-minute and FFI data, the authors demonstrate improved generalization and deliver a uniform vetting catalog of FFI TCEs, enabling more efficient follow-up and population studies. The ExoMiner++ 2.0 architecture introduces architectural refinements (merged transit-view branches, improved normalization, skip connections) and input-channel expansions (difference-image SNR, neighbor context) to robustly separate planetary signals from astrophysical and instrumental impostors, albeit with a tendency toward conservatism at stringent operating points. Overall, the approach provides a scalable, automated path to extend TESS planet vetting into the FFIs, producing actionable catalogs for the community while outlining concrete avenues for further performance gains via targeted pretraining and attention mechanisms.

Abstract

The Transiting Exoplanet Survey Satellite (TESS) Full-Frame Images (FFIs) provide photometric time series for millions of stars, enabling transit searches beyond the limited set of pre-selected 2-minute targets. However, FFIs present additional challenges for transit identification and vetting. In this work, we apply ExoMiner++ 2.0, an adaptation of the ExoMiner++ framework originally developed for TESS 2-minute data, to FFI light curves. The model is used to perform large-scale planet versus non-planet classification of Threshold Crossing Events across the sectors analyzed in this study. We construct a uniform vetting catalog of all evaluated signals and assess model performance under different observing conditions. We find that ExoMiner++ 2.0 generalizes effectively to the FFI domain, providing robust discrimination between planetary signals, astrophysical false positives, and instrumental artifacts despite the limitations inherent to longer cadence data. This work extends the applicability of ExoMiner++ to the full TESS dataset and supports future population studies and follow-up prioritization.

Figures (9)

• Figure 1: From left to right, the panels display the difference, out-of-transit, and SNR fluxes for TESS SPOC 2-min TCE TIC 83053699-1-S57 (TOI 4002.01). The neighboring stars and target's TIC-8 coordinates mapped to the CCD frame are identified as stars and as a cross, respectively. Their color encodes the stars' $T_{mag}$.
• Figure 2: Preprocessed and unnormalized neighboring stars image for TESS SPOC 2-min TCE TIC 83053699-1-S57 (TOI 4002.01). The target's TIC-8 position is shown as a red cross.
• Figure 3: Target population for the 2-min and FFI TCE datasets as a function of $T_{mag}$.
• Figure 4: High-level depiction of the new ExoMiner architecture used in this work. The details of this model are described in Section \ref{['sec:model']}.
• Figure 5: Distribution of ExoMiner++ and ExoMiner++ 2.0 scores for planet and nearby false positive TCEs.