# XTrail-ID: An Explainable AI Human Footprint Trail Identification on Soil Substrate Using Unsupervised Machine Learning from UAV Imagery

> Can a computer tell how many people walked across the ground from drone photos of their bare tracks? In Botswana, XTrail-ID grouped barefoot tracks into each person’s trail with 94% accuracy and showed heatmaps of the parts of each track it relied on. Expert trackers checked the results.

Summary by Kersey Lawrence, Original Wisdom. Updated 7 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2026-mmereki-jamisola-xtrail-id-human-tracks/

## Citation and access

- **Authors:** Wazha Mmereki, Rodrigo S. Jamisola Jr., Zoe C. Jewell, Tinao Petso, Oduetse Matsebe, Sky K. Alibhai
- **Year published:** 2026
- **Journal or source:** Machine Learning and Knowledge Extraction 8(6): 168
- **Full citation:** Mmereki, W., Jamisola, R.S., Jr., Jewell, Z.C., Petso, T., Matsebe, O., and Alibhai, S.K. 2026. XTrail-ID: An Explainable AI Human Footprint Trail Identification on Soil Substrate Using Unsupervised Machine Learning from UAV Imagery. Machine Learning and Knowledge Extraction 8(6): 168.
- **DOI:** 10.3390/make8060168
- **Access status:** Open access (CC BY)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2026_Mmereki_and_Jamisola_etal_xtrail-id-an-explainable-ai-human-footprint-trail-identification-on-soil-substrate.pdf

## Tags

- **Species:** Humans
- **Region:** Africa > Botswana
- **Tracking:** Individual identification; Tracking; Tracks; Tracks and signs identification
- **Methods and technology:** Footprint Identification Technology (FIT); Machine learning; Unmanned aerial vehicles (UAV)
- **Data type:** New field data
- **Observer qualification:** Reported but not validated
- **Discipline:** Forensics; Human tracking
- **Year published:** 2026

## Detailed keywords

- **Species:** Studied: humans, 20 healthy adult volunteers (10 men and 10 women) walking barefoot.
- **Location and habitat:** Botswana: an outdoor wildlife area with permission from the Ministry of Environment and Tourism (exact site not given); compact soft soil and loose soft soil with sparse grass, filmed in the morning and evening.
- **Topics:** human barefoot tracks on soil; grouping tracks into trails of individual people; counting the number of people in a group from their tracks; forensic, anti-poaching, military, border and search-and-rescue uses; drone imagery; self-supervised machine learning; explainable artificial intelligence and heatmaps; tracks overlapping on shared paths; expert trackers as ground truth
- **Statistics or models:** Track detection with an adapted DeepFIT model (oriented bounding boxes or segmentation masks); feature embeddings from a frozen DINOv2 vision transformer (ViT-B/14, 768 dimensions); cosine similarity, Pearson correlation and Spearman rank correlation compared; intragroup and intergroup similarity, discriminability index and grouping accuracy; paired t-tests with Bonferroni correction on 40 test images; Embedding Spatial Attribution Module (ESAM) heatmaps.
- **Data type:** New field data: 1,000 trail images extracted from 30-second drone videos flown 8 m above the ground (about 1.49 mm per pixel), 6 to 8 tracks per trail; participants walked side by side and in single file; ethics approval 10 August 2023. Inconsistencies in the paper: (1) for Figure 7c the text gives a cosine similarity of 86.6%, while the figure gives 88.6%; (2) Figure 7e is labelled “Consistency = Moderate”, while the caption describes panels e and f as low consistency; (3) section 7.6 says the oriented bounding box method overestimated the number of people by wrongly associating overlapping tracks, then repeats the paragraph saying it “merges” them, which would reduce rather than raise the count.

## Synopsis

### 1. What they did

Human tracks are valuable evidence in forensics, anti-poaching, military tracking and search and rescue, but most computer methods compare single tracks under controlled conditions. The authors developed XTrail-ID, a method that sorts barefoot tracks photographed from a drone into trails belonging to each person, without needing labelled training data.

Twenty adults walked barefoot, side by side or in single file, across compact and loose soft soil in Botswana, and 1,000 images were taken from drone video. The true trails were known from data collection and checked by an expert military tracker of the Botswana Defence Force. Tracks were located with an adapted footprint identification model, using either tilted boxes around each track or outlines of the track itself, then described by a pretrained vision model. Tracks were grouped by similarity, and heatmaps showed which parts of each track drove the grouping. Two experienced trackers checked the groupings and heatmaps.

### 2. What they found

- Using track outlines with cosine similarity gave the best results: 94.2% accuracy, against 91.5% using tilted boxes.
- Cosine similarity beat Pearson and Spearman correlations significantly for both methods.
- The outline method grouped tracks correctly for two to five people, including where grass partly hid the tracks and where people walked in single file.
- The box method worked for two or three people but became inconsistent with four or five, and overestimated the number of people when tracks overlapped on a shared path.
- The heatmaps highlighted parts of the track such as the heel, arch, outer edge and toes, which the authors interpret as differences in each person’s gait.

### 3. What they concluded

An unsupervised method using track outlines and cosine similarity can group barefoot tracks into individual trails from drone images, and heatmaps make the basis of each grouping visible, which matters in forensic work. The authors call for testing on more people, more soil types and harsher conditions, since there is no standard dataset yet.

### 4. How tracking was or was not used

Tracks are the subject. The method identifies which tracks belong to the same person and estimates how many people were present. An expert military tracker verified the true trails, and two experienced trackers independently examined the grouped trails, using track shape, trail continuity and the heatmaps, and explained why some tracks were grouped correctly or wrongly. No trailing in the sense of following a track was involved.

### 5. Tracker or observer certification

Reported but not validated. The ground truth was verified by an expert tracker specializing in military tracking in the Botswana Defence Force, and two experienced trackers evaluated the results, but no certification or test of their skill is reported.

### 6. Other technologies, skills or innovations

- Drone video (Inspired Flight IF1200A with a Sony Alpha 7R IV camera) at 8 m height.
- An adapted DeepFIT (Deep Learning Footprint Identification Technology) model, previously developed by the first author’s group for tracks on soil, used to locate each track.
- A self-supervised vision transformer (DINOv2) and a new Embedding Spatial Attribution Module to explain results with heatmaps.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- Telling how many people walked together, and which tracks belong to whom, is a classic tracking task. This study shows trackers’ knowledge being used both to check and to explain a computer method, and the method struggles most where tracks overlap on a shared path, as they often do in the field.

## Cite this summary

Lawrence, K. 2026. Summary of: XTrail-ID: An Explainable AI Human Footprint Trail Identification on Soil Substrate Using Unsupervised Machine Learning from UAV Imagery (Mmereki and Jr. et al., 2026). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2026-mmereki-jamisola-xtrail-id-human-tracks/