A reverse image search lets you search the web with a picture instead of words: you give an engine a photo, and it looks for copies of it and images that look like it. It's the fastest way to find where a photo came from, whether it has been edited, and whether the person in a profile picture is who they say they are. Below: how it works, step-by-step instructions for the four major free engines, and the limits and privacy trade-offs.
Key takeaways
- Engines don't compare pixels. They turn your image into a compact "fingerprint" and look for close fingerprints in an index they built by crawling the web.
- Each engine has its own index and its own idea of a "match," so running the same photo through two or three engines finds far more than one.
- TinEye is best for exact and edited copies and for finding the earliest copy it knows about; Google Lens is best for broad and visually similar matches; Bing is handy for cropping to one object; Yandex is useful for non-English sites.
- No results does not prove a photo is original. Private accounts, many social networks and brand-new uploads are often not indexed.
- Uploads are handled differently: TinEye says it deletes them, Google can save them if you've turned on Visual Search History, and Microsoft says Bing may use them to improve its services.
Disclosure: this page contains an affiliate link. If you sign up through it, we may earn a commission at no extra cost to you. How we work.
What reverse image search is (and isn't)
A normal search starts with words and returns pages. A reverse image search starts with an image (an upload, a pasted image, or an image URL) and returns pages that contain the same or a similar image. People use it to:
- Check a person's photos. If a dating match's pictures show up on a model's portfolio under a different name, that's a red flag. The FTC's own romance-scam advice tells you to do a reverse image search of the person's profile picture. Our catfish check guide covers the full process.
- Trace a viral or news photo back to its earliest known copy to see whether it's old, out of context or altered. See how to find the original source of an image.
- Find copies of your own work used without permission so you can request removal.
- Vet listings such as rentals or marketplace items whose photos may be lifted from elsewhere, or find a larger version of a low-resolution image.
It overlaps with, but isn't the same as, visual search. Visual search tries to understand what's in the picture ("this is a monstera plant," "this lamp costs $40 here"). Reverse image search is about where this particular picture appears. Google Lens and Bing do both at once, which is why their results mix shopping links with source pages.
How reverse image search works
No engine compares your photo pixel by pixel against every image online; that would be too slow and would fail whenever an image was resized. Instead, engines follow roughly the same three stages.
1. Building the index
Long before you search, a crawler visits public web pages and stores a compact description of each image it finds, along with the page URL and the date found. This is the index. An engine can only find what it has crawled, which is why engines return different results.
2. Feature extraction: the image "fingerprint"
When you search, the engine runs your image through the same process it used on the indexed images. There are two broad families of technique:
- Perceptual hashing. The image is shrunk, converted to grayscale and reduced to a short string of bits that captures its overall structure (where it's light and dark, how brightness changes). Unlike a cryptographic hash, a small edit changes only a few bits, so two copies of the same photo at different sizes produce nearly identical hashes. Similarity is measured by counting differing bits. This approach is good at finding copies.
- Learned features (embeddings). Modern systems also pass the image through a neural network that outputs a list of numbers describing its content: shapes, textures, objects, scenes. Two different photos of the same landmark or a similar-looking product land close together in this "feature space." This approach is good at finding similar images, not just copies.
Engines don't publish their exact recipes, but the trade-off is consistent: fingerprints tuned for copies are precise but miss look-alikes; fingerprints tuned for similarity cast a wider net but return more noise.
3. Matching and ranking
The engine then looks for indexed fingerprints closest to yours. With billions of entries it can't check every one, so it uses approximate nearest-neighbor indexes that jump quickly to the right neighborhood. Results are ranked by similarity and then, depending on the engine, re-sorted or grouped by things like page relevance, image size or date found.
How to use Google Lens on a computer
Google's reverse image search now runs through Google Lens. You can start it from the Google homepage in any major browser, or from the right-click menu in Chrome.
From google.com (any browser)
- Go to google.com (or images.google.com) and click the camera icon, labeled Search by image, in the search box.
- Either drag an image file into the box, click Upload a file and choose it, or paste a web address into Paste image link and click Search. To get an address, right-click an image on a website and choose Copy Image Address (wording varies slightly by browser).
- Results open in Lens. Your image appears on one side with an adjustable selection box; drag its corners to focus on one face, object or area and the results update.
- Look through the tabs above the results. Google labels and orders them differently over time, but look for an Exact matches option when you want copies of the same picture rather than look-alikes.
From Chrome's right-click menu
- On any web page in Chrome, right-click an image.
- Choose Search with Google Lens. Results open in a side panel next to the page, so you don't lose your place.
- Adjust the selection or open the results in a full tab for more room.
Lens is the strongest choice for a first, broad search because Google's web index is so large and it matches visually similar images, not just copies. The downside is noise: a photo of a person may return dozens of strangers with similar hair and clothing, and a photo of an object often leads with shopping results. In some regions Google also offers an About this image panel showing when Google may first have seen an image, which helps when dating a photo. On a phone, see reverse image search on iPhone and Android.
How to use TinEye
TinEye, run by the Toronto company Idée since 2008, keeps its own index of tens of billions of images and is built for one job: finding copies of a specific image, including cropped, resized, recolored and lightly edited ones. It does not do facial recognition and rarely returns merely similar pictures.
- Go to tineye.com. Click the upload button, drag an image onto the page, or paste an image or image URL into the search box.
- Check that the file fits: TinEye accepts common formats including JPEG, PNG, GIF, BMP and TIFF up to 20 MB, and works best with images at least 300 pixels on a side.
- Use the Sort menu. Best Match is the default; Most Changed shows heavily edited versions first; Biggest Image finds the highest-resolution copy; Newest and Oldest sort by when TinEye found each copy.
- Narrow results with filters. The stock filter limits results to stock-photo sites, and Filter by website restricts results to one domain.
- Click a result's thumbnail to open the compare view, which flips between your image and the match so you can spot crops and edits.
Tip: TinEye has browser extensions for Chrome, Firefox, Edge and Opera that add a right-click "search on TinEye" option. If you check images often, they save the download-and-upload step.
One caution about Oldest: it shows the earliest copy TinEye found, which is not necessarily the first time the photo was published. Treat it as a strong lead, then confirm on the page itself.
How to use Bing Visual Search
- Go to Bing and select the Images tab, then click the camera icon in the search box.
- Drag in an image, paste an image or URL, browse for a file, or take a photo with your webcam.
- On the results page, resize the selection box around the part of the image you care about. This is Bing's most useful feature: you can isolate one object in a busy scene.
- Scroll for the section listing pages that include the image, plus related images and, for products, shopping matches.
In Microsoft Edge, the same search is available from an image's right-click menu. Bing is a good second opinion after Google, especially for products and landmarks.
How to use Yandex Images
- Go to yandex.com/images and click the image search button to the right of the search bar.
- Drag the image onto the upload panel, click Select file, paste from your clipboard, or enter an image address and click Search.
- Review the results: other sizes of the image, sites that contain it, and similar images. Yandex can also recognize and translate text in the picture.
Investigators value Yandex for its coverage of Russian- and Eastern European-language sites and for returning look-alike faces, though some report its coverage of Western social media has thinned in recent years. Yandex is headquartered in Russia; keep that in mind before uploading anything sensitive.
How to read your results
Results fall into three types, and confusing them is the most common mistake:
| Match type | What it means | What to do with it |
|---|---|---|
| Exact copy | The same photo, possibly resized or recompressed | Compare names, captions and dates on each page |
| Modified copy | Cropped, mirrored, filtered, watermarked or text added | Look for the least-edited, largest version; it's usually closest to the source |
| Visually similar | A different photo that looks alike | Useful for identifying things; not evidence that two people are the same |
When you find copies, ask four questions:
- Which copy is oldest? Use TinEye's Oldest sort and Google's About this image, then check the dates on the pages themselves.
- Which is largest? Originals are usually higher resolution than reposts.
- Do the names match? The same face under three names is a classic sign of a fake profile.
- Is there a credit or watermark? These often name the creator.
If the photo is of a person you're talking to and the free engines turn up nothing, keep in mind that many dating and social sites aren't indexed by public engines. Paid people-search services such as Social Catfish search some of those sources and cross-reference contact details. Weigh the cost against what you need, and compare options in our tools comparison.
Tips when you get no matches
Reused images are often altered, sometimes deliberately to defeat reverse image search. Before concluding a photo is unique:
- Crop to the subject. Remove app borders, captions, emoji and stickers. Screenshots of a phone screen add a lot of clutter that pulls the fingerprint away from the original.
- Flip it. Mirroring is a common trick. Flip the image horizontally in any photo editor and search again.
- Search the best version you have. Use the original file or the largest copy rather than a blurry screenshot.
- Search parts separately. A face, a background landmark and a distinctive object may each match different sources.
- Use the text. Signs, shop names and logos can be searched as ordinary keywords.
- Try another engine. See our TinEye alternatives for what each covers. For video, search a clear still frame.
Limitations to keep in mind
- Only indexed, public images can match. Private profiles, closed groups, messaging apps and many dating sites are invisible to public engines. TinEye says plainly that it doesn't index social media sites such as Facebook or LinkedIn.
- New images lag, and dates are discovery dates: "oldest" means oldest found by that engine.
- Heavy edits break matching. Big crops, collages, strong filters and redrawn images can fall outside what an engine treats as a copy.
- Similar is not the same. A visually similar face is not the same person. Never accuse anyone based on a look-alike result.
- AI-generated faces have no source. A synthetic profile photo won't match anything because it never existed before.
What happens to the photos you upload
Uploading a photo means handing a copy to a company. Policies differ a lot:
- TinEye says it does not save the images you upload, does not add them to its index, and deletes them after searching.
- Google lets you choose whether images you search with Lens are saved to your account through a Visual Search History setting, which Google says is off by default. When it's on, those images are stored in your Web & App Activity, and you can delete them at activity.google.com.
- Bing: Microsoft's help page notes that photos you provide may be used to improve Bing's image-processing services.
- Yandex's image-search help doesn't describe how uploads are stored, so assume they may be retained.
Before you upload: don't search with images of ID documents, intimate photos or children. If the image is already public, paste its URL instead of uploading a file. Camera files can also carry location data in their metadata.
Frequently asked questions
Is reverse image search free?
Yes. Google Lens, TinEye, Bing Visual Search and Yandex Images are free for personal searches. Paid tools mostly add face recognition or people-search records.
Can reverse image search find a person from a photo?
It can find pages where the same photo appears, which often reveals a name. General engines are not designed to identify faces. Dedicated face search engines exist but raise serious privacy and ethical issues. None of the four engines here notifies the person in the photo.
Why do Google and TinEye give different results?
They crawl different parts of the web and define a "match" differently. TinEye looks for copies of your image; Google also returns visually similar images and product matches. Running both is normal practice.
What does "no results" mean?
Only that the engine has no indexed copy it recognizes. The image may be on private or unindexed sites, may be too heavily edited, or may be newly created, including by AI.
Sources & further reading
- Google Search Help: Search with an image on Google
- Google Search Help: Visual Search History
- TinEye: How to use TinEye to search for an image
- Microsoft Support: Using Bing Visual Search
- Yandex Images Help: Search by image
- FTC Consumer Advice: What to know about romance scams
Spotted an error or something out of date? Tell us — we review corrections promptly. Read how we research and update guides in our editorial policy.