Bounding-box image annotation for Mac
Draw a labelled box around every object in a folder of images, then take the whole set out as one CSV, ready for training.
Step 1 of 4
Drag a box around the fruit, type its class or click one you have used, and Command-N brings up the next photo. The four boxes on this plate:
Step 2 of 4
Every photo you move on from gets a tick in the sidebar, and a blue ring marks the one you are on, so a long set never loses its place.
Step 3 of 4
Name a class once and it turns into a button, with its own colour on every photo. Apple is spelled the same way on the first image and the last.
Step 4 of 4
The share button saves annotations.csv: a row per image, with every box’s label and its x, y, width and height in pixels.
Cinder
Bilbo
Haplo AI
Barrier
ShipHappens
Haplo AI Investing
ScriptCam
Markdown2PDF
EasyLogo
Annot8Filename,Label 1,CGRect 1,Label 2,CGRect 2,Label 3,CGRect 3,Label 4,CGRect 4,Label 5,CGRect 5,Label 6,CGRect 6,Label 7,CGRect 7,Label 8,CGRect 8,Label 9,CGRect 9,Label 10,CGRect 10,
"plate-apples-orange-banana.jpg","apple","837.8, 179.5, 699.0, 720.8","apple","1740.8, 68.0, 671.8, 718.1","orange","1335.5, 862.2, 794.2, 780.6","banana","750.7, 1357.3, 1912.2, 911.2",,,,,,,
"table-bananas-oranges.jpg","orange","408.6, 723.6, 1200.4, 1332.3","orange","3486.2, 404.4, 1140.8, 1132.3","orange","3665.0, 1553.7, 1225.9, 1183.4","orange","800.3, 2213.5, 1242.9, 1196.1","banana","1600.5, 178.8, 766.2, 2409.3","banana","1826.1, 25.5, 1034.4, 2703.0","banana","2434.8, 0.0, 936.5, 2788.1","banana","3009.5, 366.1, 821.5, 2890.3","banana","2519.9, 2656.2, 638.5, 578.9",,
"bowl-oranges-limes-apple.jpg","orange","29.9, 603.7, 736.0, 672.0","orange","765.9, 712.5, 507.7, 364.8","orange","535.5, 1297.1, 605.9, 631.5","apple","1124.3, 1254.4, 691.2, 689.1",,,,,,,
"book-three-apples.jpg","apple","1975.3, 1413.8, 1037.8, 1102.9","apple","0.0, 2727.3, 1238.3, 1163.1","apple","1183.1, 3830.2, 1539.1, 1664.4",,,,,,,,
"empty-produce-baskets.jpg",,,,,,,,,,,
"apples-pears-lemons.jpg","apple","777.5, 489.3, 1211.9, 1125.0","apple","3827.9, 214.9, 1202.8, 1120.5","apple","1980.3, 1436.0, 1184.5, 1143.3","apple","4010.8, 2332.4, 1175.3, 1225.7",,,,,,,
"nuts-apples-oranges.jpg","orange","248.1, 24.8, 138.9, 129.9","orange","177.8, 112.5, 124.9, 129.4","apple","134.4, 255.9, 117.0, 117.6","apple","228.9, 215.4, 94.5, 97.9","apple","299.8, 146.8, 99.6, 95.1","apple","397.1, 126.0, 91.1, 78.2","apple","324.6, 222.8, 89.4, 81.0","apple","412.3, 193.5, 90.0, 75.4","apple","271.7, 279.6, 114.2, 104.6","apple","379.1, 249.8, 120.4, 110.8",
The set’s 34 boxes in Annot8’s CSV layout, measured on the photos. Turn it into YOLO labels with our guide
Photos from Wikimedia Commons: Happy fruits by Tristan Schmurr (CC BY 2.0); A group of ripe bananas sits next to several oranges by Shixart1985 (CC BY 2.0); Fruits in bowl: oranges, lime, apples by Gerardolagunes (CC0); Three apples, red and green, rest near an open book on a wooden table by Shixart1985 (CC BY 2.0); Empty pint-sized molded pulp produce baskets by Kayaker (CC BY-SA 4.0); Fresh fruits arranged on a wooden surface by Shixart1985 (CC BY 2.0); Nuts, apples and oranges by Zeynel Cebeci (CC BY-SA 4.0). Resized, with our boxes drawn on them; the CC BY-SA photos are shared under the same licence.
Annotation is a numbers game: the tool should never be the slow part. Draw, name, next.
Drag a box, pick its label, then Command-N takes you to the next image. The loop stays under your hands. Here, the nine boxes on our bananas-and-oranges photo.
Command-N or the right arrow moves through the set and the canvas stays put. Return saves a label, and nothing waits on a web app to catch up with your hands.
A class you have typed once comes back as a one-click button, in the same colour on every image, so it never drifts between photos.
The sidebar ticks off every image you have been through, so you always know how far into a set you are.
One row per image, with every box’s label and coordinates, ready to reshape into whatever your training pipeline expects. Our guide turns it into YOLO.
No upload, no seat licence, no per-image pricing. Useful when the images are medical, proprietary, or simply not yours to put on someone’s server.
One payment, against annotation platforms that bill by the month or the seat. Label as many sets as you like on your Mac.
Every preview and screenshot, straight from the App Store listing.
$1.99 once, for Mac. Open a folder, draw your boxes and save the CSV. No account, and no subscription.
Annot8
Image annotation for Mac
.jpg){kind=link}
