Abstract:Introduction: Habitats with a taphonomic window are high priority targets for astrobiology missions to Mars [1]. However, if life existed on Mars it did not interact with its physical environment in a way that has been obvious to identify from orbit. Studying terrestrial Mars analogs can inform us of the types of features that serve as habitats with a taphonomic window on Earth, and by extension the types of features that should be sought on Mars. But what spatial resolutions are necessary to positively identify such targets? To address the problem of establishing thresholds of identification in spatial resolution, we are assessing the ability of a deep convolutional neural network (CNN) to perform semantic segmentation (classify each pixel) on images of the astrobiologically-relevant Mars-analog environment, Salar de Pajonales (SdP), in Chile (Fig. 1A). Our aim is to discover the resolutions at which the habitats at SdP are no longer identifiable by the CNN to understand if the same types of features could be identified in salt-encrusted basins on Mars. CNNs excel at many tasks related to image analysis, including image classification [2], with modern CNNs capable of outperforming human analysts [2, 3]. Our investigation requires a consistent analysis of the same terrain over many ground sampling distances (GSDs). A single human analyst would become biased after the first classification task, and many human analysts would vary in their ability to classify a scene. Therefore, a CNN is a natural choice to tackle this problem. Here we present initial results of the semantic segmentation of our highest resolution (3 cm/pixel) scene of SdP. In future work, we will evaluate the CNN performance at progressively lower GSDs to establish thresholds of identification for habitats at SdP. Geologic Background and Relevance to Mars: Salar de Pajonales (SdP) is a salt-encrusted basin in the Chilean Altiplano (25°08'29"S, 68°46'20"W, 3547m, Fig. 1A) [4]. Its polyextreme characteristics make it a suitable analog for a post-Noachian martian climate [5]. Our study site within SdP is a gypsum-dominated area that notably hosts ridges and tumuli [6] that serve as endolithic habitats [5], and a “patterned ground” unit that represents a biological soil crust [5] (Table 1, Fig. 1B). Data and Methods: We captured images of our

A Global Database of Pitted Cones on Mars for Research on Martian Volcanism

Mapping Cones on Mars in High-Resolution Planetary Images with Deep Learning-Based Instance Segmentation

Analyzing spatial distributions and alignments of pitted cone features in Utopia Planitia on Mars

ConeQuest: A Benchmark for Cone Segmentation on Mars

GMSRI: A Texture-Based Martian Surface Rock Image Dataset

Mapping "Brain Terrain" Regions on Mars using Deep Learning

CONVOLUTIONAL NEURAL NETWORK IMAGE CLASSIFICATION OF MARS-ANALOG TERRAIN: PRELIMINARY RESULTS AND IMPLICATIONS FOR THE SEARCH FOR LIFE ON MARS.

Automated detection of geological landforms on Mars using Convolutional Neural Networks

DoMars16k: A Diverse Dataset for Weakly Supervised Geomorphologic Analysis on Mars

High-Precision Measurement of 3-D Rock Morphology on Mars Using Stereo Rover Imagery

Segmentation Convolutional Neural Networks for Automatic Crater Detection on Mars

Planet Four: A Neural Network's Search For Polar Spring-time Fans On Mars

Analysing high resolution digital Mars images using machine learning

AI4MARS: A Dataset for Terrain-Aware Autonomous Driving on Mars

An automatic classification method for mapping Martian landforms

Surveying the ice condensation period at southern polar Mars using a CNN

Joint Hapke Model and Spatial Adaptive Sparse Representation with Iterative Background Purification for Martian Serpentine Detection

Automated Rock Detection From Mars Rover Image via Y-Shaped Dual-Task Network With Depth-Aware Spatial Attention Mechanism

A Novel Method to Extract Rocks from Mars Images

Massive Stereo-Based DTM Production for Mars on Cloud Computers

Deep Mars: CNN Classification of Mars Imagery for the PDS Imaging Atlas