The detection will handle faces occluded up to approximately 50%, given that the eyes are not occluded. The size of the face in the image should be at least 40x40 pixels. The system can be trained to handle such poses in the future. crawling, lying down) are currently not handled. In cases where the head and shoulders of the person are occluded, detection might not be possible. The height of the person in the image should be at least 15% of image height.įor detecting people, the head and shoulders of the person are important. Detections may not trigger if the scene is too dark.Īngle of the camera to the object being detected should be within 45 degrees for best results. Good lighting conditions are important for detection. The blurrier the object the more difficult to detect. Paid accounts (Basic, Pro, and Enterprise) have access to separate, production-ready servers.īlurry scenes (possibly caused by camera settings or weather conditions) will reduce the accuracy of the object detector. All Cloud accounts can use the free, development service for up to 5,000 API calls per month. Select the appropriate base URL for your environment and Cloud API account level. Code samples for a variety of languages are available on the Sighthound Cloud demo page. Faces can be analyzed for gender, age, pose, or emotion and a landmark detector can find the various facial features in the detected faces, including eyes, nose and mouth, by fitting 68 landmark points to those features.īefore proceeding, please ensure you have created a Sighthound Cloud API Token. The Detection API returns the location of any people and faces found in an image.
0 kommentar(er)
