Count Faces in a Photo — AI Face and People Counter
Works with JPG, PNG, WEBP
People search for a face counter when what they want is a people count — and faces alone undercount every scene with someone turned away. Upload a photo of the photo and get a headcount with every person marked for review.
- Counts dense crowds as well as small groups
- Built for anyone counting a photo
- Stop recounting because someone moved halfway through
- Review every detected person directly on the marked result
Works with JPG, PNG, WEBP
Drop a photo of the photo — get back a headcount with every person marked
Tips for an accurate count
• If some people show only the backs of their heads, switch on “include people facing away” — faces are not required for the count.
• Raise sensitivity if some people are missed in small groups
• If people are turned away or in profile, turn on “Include people facing away” — it counts them too, not just visible faces; check the markers for any duplicates
About the People Counter
People search for a face counter when what they want is a people count — and faces alone undercount every scene with someone turned away. The people counter picks the right engine for the photo: small groups are detected person by person with verifiable numbered boxes, while dense scenes switch to an AI crowd-density engine built for gatherings where bodies overlap and detectors undercount. If some people show only the backs of their heads, switch on “include people facing away” — faces are not required for the count. Review the marked result, correct anything it missed, then download the annotated image as your record.
How It Works
1. Upload a photo of the group or crowd
2. AI detects each person
3. Adjust sensitivity to catch more or fewer
4. Read the headcount and download the marked image
Perfect For
• Face-visible headcounts of posed groups
• Counts that include people facing away
• Quick checks that everyone made it into frame
• Marked photo records you can file or share
Face counter or people counter? Counting heads in a photo
People search for a face counter when what they usually want is a people count. Faces alone undercount any scene where someone is turned away, wearing a mask, looking down at a phone or standing behind someone else. With the facing-away option on, this page counts people whether or not their face is visible.
In small groups the counter puts a numbered box on each person it finds. Turn on "Include people facing away" when some people show only the backs of their heads or stand in profile; it adds whole-body cues so a visible face isn't required, and "Max accuracy (slower)" combines results from more than one counting method where available. Overlap and partial bodies can still cause misses or a duplicate marker, so check the numbered boxes. It doesn't recognize or name anyone; it only marks where a person is.
If you need the opposite, hiding faces before sharing a photo, pixelate or blur faces using its face-detection option. Posed group shots are covered on the group photo page, and the main people counter has every option for everyday headcounts.
Before counting heads
- Turn on the facing-away option for turned heads
- Use the largest copy of the photo
- Check the back rows for hidden faces
- Correct the total for missed or doubled boxes
Set up for Count Faces in a Photo — AI Face and People Counter
Prepared settings
- Starting point
- One wide photo of the photo
- If some people show only the backs of their heads, switch on “include people facing away” — faces are not required for the count.
- Prepared workflow
- Count faces
- People mode and automatic scene routing are already selected; there is no detector to find manually.
- Reviewable output
- A marked headcount and downloadable record
- Detected people are marked so the total can be checked against the source instead of accepted as a black box.
- Review rule
- Review every marker before using or sharing the total
- Occlusion, blur, and people at the frame edge can affect a count; correct any missed or duplicate marker first.
Best-fit uses
- Face-visible headcounts of posed groups
- Counts that include people facing away
- Quick checks that everyone made it into frame
Frequently asked questions
- How do I count faces from a photo?
- Upload one clear photo of the photo and the AI marks every person it detects, with the total shown on top. If some people show only the backs of their heads, switch on “include people facing away” — faces are not required for the count. Review the markers and adjust sensitivity before you rely on the number.
- Does it work when the scene is packed?
- Yes — dense scenes switch automatically to a crowd-density engine built for gatherings where people overlap, with a heatmap and dots showing where the estimate comes from. Occlusion still hides people, so treat busy-scene results as reviewed estimates and use a higher, wider vantage when you can.
- Is counting faces different from counting people?
- Yes, and it matters: a face counter misses everyone turned away from camera. Turn on “Include people facing away” and the tool also counts people whose face is hidden, using whole-person cues — which is almost always the number you actually wanted. Check the numbered markers for anyone missed or marked twice.
- Is the people counter free, and what happens to my photo?
- Yes — count, review and download the marked-up result without creating an account, and the headcount itself is identical on every plan. Downloads are free without a watermark. No signup required. Photos are encrypted in transit, processed to produce the count, and removed under our Security policy.
What you can do with People Counter
Review the main capabilities before you begin.
- Count people in a crowd without losing your place
- Estimate event attendance from a single image
- Tally heads in classrooms and gatherings
- See each counted person marked on the image
- Include people facing away from the camera
- Turn on max accuracy for dense crowds
- Fine-tune detection for small or distant faces
Settings reference
6 settings
Available controls for People Counter.
- Processing modePresets
- Process one input, or switch to Premium batch mode to process multiple inputs with shared settings. Available input types depend on the tool.Options: Single photoBulk audit
Count
- Included markersSliderRange: 0–1 · Default: 0.21
- Starts at 21%; slide toward Show more to include weaker people markers or Only strongest to hide them. The headcount updates instantly.
- Overlay stylePresets
- Sets how counted people appear on the preview — dots, a density heatmap, both, or a clean image — so you can verify the headcount visually.Options: BothDotsHeatNone
- Include people facing awayOn / Off
- Counts people turned away, in profile, or whose face is hidden — not just visible faces — while never counting one person twice and never counting legs alone as a person. For genuinely dense crowds it escalates to a density estimate.
- Max accuracy (slower)On / Off
- Runs every counting engine on the photo and cross-checks their results for the most dependable headcount, taking a little longer to process.
- Detection sensitivitySliderRange: 0–1 · Default: 0.5
- Controls how eagerly people are marked, from Strict to Lenient — raise it if people in small groups are missed, lower it if things that are not people get counted.
Popular presets
Related tools for Count Faces in a Photo — AI Face and People Counter
Tools for common next steps after Count Faces in a Photo — AI Face and People Counter.
Counting & Inventory
Object Counter
Estimate an object count from a photo. Review and correct the numbered markers before using the count.
Try it now
Counting & Inventory
Live Object Counter
Aim your phone at separated, fully visible items and review the detector markers as the count updates in real time.
Try it now
OCR & Data Extraction
Image to Text (OCR)
Extract editable text from a screenshot, photo or scan in 30+ supported languages.
Try it nowHow People Counter helps you get it done
Real problems it solves every day — for businesses, creators, and everyday tasks. Find the use case that fits you and start.
Event & Conference Attendance Reconciliation
Photograph the room or venue for a preliminary attendance estimate, then review the result and use an approved manual or ticketing count for reconciliation.
Crowd-Size Estimates for Rallies & Protests
Generate a rough crowd-size estimate from an overhead or wide photo, review the uncertainty, and corroborate it with other evidence before publishing or making decisions.
Retail Footfall & Queue Snapshots
Count shoppers in a queue or at a storefront from one photo to spot-check footfall or queue length at a specific moment, with no camera system required.
Classroom & Meeting Attendance Registers
Use one room photo for a preliminary head-count estimate, then have an authorised person complete or reconcile the attendance register manually.
Preliminary Venue Occupancy Estimates
Use a photo estimate only as an early warning, then complete the approved physical or official occupancy check before opening. The AI result is not a fire-safety or capacity-compliance record.
Transit Platform & Waiting-Area Crowd Checks
Photograph a train platform, bus stop or waiting area for a preliminary density estimate, then corroborate it with approved operational checks before making safety or capacity decisions.
Guides for this tool
- A Busy Café Photo Tells One Part of the Demand StoryA busy cafe photo captures occupancy at one moment. Explore Pixoate people counts, their limits and how reviewed snapshots can inform demand discussions.
- Count Restaurant Customers from CCTV Photos: A Practical GuideUse café and restaurant photos to review occupancy and queues. Learn a checked headcount workflow and why snapshots cannot measure unique daily visitors.
Watch each transformation land
Every photo crosses the slash untouched and comes out the other side finished.
