{{ title }}
Cameras
{% if issues %}{{ issues|length }} of {{ deployments }} deployments had problems that may affect the results:
-
{% for name, found in issues.items() %}
- {{ name }}: {{ found|join("; ") }} {% endfor %}
✓ No camera problems found in {{ deployments }} deployments.
{% endif %}Each tick is an image; grey bars are deployment periods; shaded areas are long gaps without images.
Species
{% if species_rows %}| Species | Events | RAI | Sites | Naive occ. | Occupancy | {% if ai_errors %}Sites in doubt | {% endif %}Busiest |
|---|---|---|---|---|---|---|---|
| {{ row.name }} | {{ row.events }} | {{ row.rai }} | {{ row.sites }} | {{ row.naive }} | {{ row.modelled }}{% if row.interval %}{{ row.interval }}{% endif %}{% if row.warning %}⚠ unreliable{% endif %} | {% if ai_errors %}{{ row.doubt }}{% if row.doubtful %}⚠ review{% endif %} | {% endif %}{{ row.peak }} |
Events are independent visits (a new event after 30 minutes without images at a camera). RAI is events per 100 camera-days. Sites: camera sites where the species was recorded, of all sites. Naive occupancy is that share; occupancy corrects for the species being missed where it is present, with its 95% interval below, using {{ occasion_days }}-day sampling occasions, for species recorded at three or more sites. ⚠ marks estimates that are unreliable because the species was recorded too rarely (hover for the reason). Busiest is the time of day with most activity, for species with {{ min_activity_events }} or more events.{% if ai_errors %} Sites in doubt: the expected number of sites with a false record / where the species was missed because of AI mistakes, with the AI's confidence calibrated on {{ labels.reviewed }} reviewed events; ⚠ marks species whose results may change with more review.{% endif %}
{% if left_out %}Events identified only above species level are not in the table: {% for name, count in left_out.items() %}{{ name }} ({{ count }}){% if not loop.last %}, {% endif %}{% endfor %}.
{% endif %} {% else %}No animal events identified to species.
{% endif %}Daily activity
{% if activity_chart %}Overlap between species' daily activity (0 = never active at the same time, 1 = identical):
| Species | Overlap (95% interval) |
|---|---|
| {{ pair.first }} and {{ pair.second }} | {{ "%.2f"|format(pair.overlap) }} ({{ "%.2f"|format(pair.overlapLower) }}–{{ "%.2f"|format(pair.overlapUpper) }}) |
No species has {{ min_activity_events }} or more events, the fewest for an activity curve.
{% endif %}Species richness
{% if richness %}The survey recorded {{ richness.observedSpecies }} species in {{ richness.cameraDays }} camera-days. The Chao2 estimate of the species present is {{ richness.estimatedSpecies }} (95% interval {{ richness.estimatedLower }}–{{ richness.estimatedUpper }}), so the survey found about {{ "%.0f"|format(richness.completeness * 100) }}% of them. {{ richness.uniques }} species were recorded on only one camera-day.{% if not labels.reviewed %} Some of those may be AI mistakes rather than rare species; reviewing them makes this estimate more reliable.{% endif %}
No species recorded.
{% endif %}How the species were identified
Animals were found and identified by {{ labels.classifiers|join(", ") if labels.classifiers else "AI" }}. {% if labels.reviewed %}A person reviewed {{ labels.reviewed }} of {{ labels.events }} events ({{ "%.0f"|format(100 * labels.reviewed / labels.events) }}%), starting with those the AI was least sure about; their decisions replace the AI's.{% else %}No events were reviewed by a person, so all labels are the AI's.{% endif %}
{% if labels.accuracy %}Compared with reference labels for {{ labels.accuracy.events }} events, the labels were correct for {{ labels.accuracy.correct }}, correct only above species level for {{ labels.accuracy.higher }},{% if labels.accuracy.finer %} more specific than the reference (which could not confirm the species) for {{ labels.accuracy.finer }},{% endif %} and wrong for {{ labels.accuracy.wrong }}.
{% endif %}Methods and limitations
- RAI compares how often species are recorded; it is not a measure of density, because species differ in how easily cameras record them.
- Occupancy estimates with a warning are unreliable: the species was recorded too rarely, or at too few sites, to separate rarity from being hard to detect.
- AI labels contain mistakes, especially for night images and rare species; species never recorded may be present.
- Detection histories, abundance indices, activity curves and the full model results are produced by
camtraptools ecology.