OPENCV_PERFORMANCE(1)

💖 OPENCV_PERFORMANCE 💖

okay bestie, gather round, because this little tool — opencv_performance from the OpenCV user commands — is literally the vibe check for your classifier. it's how you find out if your object detector is slaying or flopping. no cap. 🦋

---

✨ NAME ✨

opencv_performance — evaluate the performance of the classifier 👑

---

🎀 SYNOPSIS 🎀

opencv_performance [options]

---

🌸 DESCRIPTION 🌸

so here's the tea: opencv_performance evaluates the performance of the classifier. you hand it a whole collection of marked up test images, it runs the classifier on them, and then it spills all the results — number of found objects, number of missed objects, number of false alarms, and other information. it's giving report card energy. 📋

if you don't have a collection like that lying around, no worries bestie — test samples may be created from a single object image by the opencv_createsamples(1) utility. and the scheme of test samples creation in this case is similar to training samples. same vibe, different day. ✨

now, when you read the output, here's how to decode that table:

• 'Hits' — shows the number of correctly found objects. we love to see it. 💅
• 'Missed' — shows the number of missed objects (must exist but are not found, also known as false negatives). the audacity. 😭
• 'False' — shows the number of false alarms (must not exist but are found, also known as false positives). not the phantom detections 💀

---

💅 OPTIONS 💅

opencv_performance supports the following options:

• -data classifier_directory_name — the directory, in which the classifier can be found. point me to the classifier, bestie 📁
• -info collection_file_name — file with test samples description.
• -maxSizeDiff max_size_difference — determine the size criterion of reference and detected coincidence. the default is 1.500000.
• -maxPosDiff max_position_difference — determine the position criterion of reference and detected coincidence. the default is 0.300000.
• -sf scale_factor — scale the detection window in each iteration. the default is 1.200000. 🔍
• -ni — don't save detection result to an image. this could be useful, if collection_file_name contains paths. lowkey a lifesaver 🙈
• -nos number_of_stages — number of stages to use. the default is -1 (all stages are used). go big or go home, all stages periodt ✨
• -rs roc_size — the default is 40.
• -h sample_height — the sample height (must have the same value as used during creation). the default is 24.
• -w sample_width — the sample width (must have the same value as used during creation). the default is 24.

heads up bestie: the same information is shown, if opencv_performance is called without any arguments/options. so if you just run it bare, it'll still give you the rundown 🫶

---

🌈 EXAMPLES 🌈

to create training samples from one image applying distortions and show the results:

opencv_performance -data trainout -info tests.dat

that's it. that's the whole thing. iconic behavior honestly 🎀

---

👑 SEE ALSO 👑

opencv_createsamples(1), opencv_haartraing(1)

more information and examples can be found in the OpenCV documentation. iykyk 📚

---

💜 AUTHORS 💜

this manual page was written by Daniel Leidert <daniel.leidert@wgdd.de> and Nobuhiro Iwamatsu <iwamatsu@debian.org> for the Debian project (but may be used by others). absolute legends, we stan 🫶

---

and that's the whole page, bestie! this masterpiece dropped under the OpenCV name, dated May 2010 — vintage but still serving 💖✨🌸 bye pookie, go test those classifiers 🦋