With a case catalog! Quickly detect human faces and both eyes from natural images! Utilized in a wide range of fields!
Our company's face detection processing allows for easy face registration using commercially available web cameras. Multiple users can also be easily registered, making it suitable for entry management and other applications. Recognition is possible if the distance between the eyes is 24 pixels or more (32 pixels or more is recommended), and even small faces can be authenticated. 【Features】 ● In speed priority mode, face detection from VGA input images takes about 30ms (depending on minimum detection size) on an Intel Core i7 2.93GHz. ● Easy to port to embedded versions due to high-speed processing and small footprint. ● Detection is based on eye features rather than estimation from face position, allowing for detection even when wearing glasses or masks. For more details, please contact us or download our catalog to view.
Inquire About This Product
basic information
It determines and detects whether there is a face in the image. The tilt of the face and the minimum detection size can be set as parameters. When prioritizing speed, it quickly detects faces with a tilt of ±45 degrees (tilt from upright). When prioritizing accuracy, it detects faces in any orientation with high precision, up to 360 degrees. Additionally, the detection angle for the left and right direction of the face can be detected up to ±30 degrees. The minimum detection size is 24 pixels or more between the eyes. The number of detections can be adjusted, but by default, it is almost unlimited. Furthermore, the basic condition for detection assumes that the eyes, eyebrows, mouth, and nose are visible.
Price range
Delivery Time
Applications/Examples of results
As part of embedded support, there is also a track record of porting to TI DSP (C674x). Processing is possible in 50ms at a clock speed of 456MHz. Reference specifications: Input image size QVGA (320×240), detectable minimum size: 20×20.
catalog（1）
Download All Catalogs
Company information
The 21st century is an era of intense competition in science and technology. For companies, there is an urgent need to transition from mass production with high quality and low cost to the development of "Only One" products with differentiated technologies. In response to this background, TakumiVision was established as a technology development venture originating from Ritsumeikan University. In collaboration with Ritsumeikan University and Shiga University of Medical Science, we develop highly original proprietary algorithm libraries in the fields of image processing, audio processing, and medical diagnostics, which we provide as intellectual property. We are also skilled in implementing algorithms into LSI and embedded systems, considering chip implementation during the algorithm development stage and accommodating custom requests for implementation. Additionally, we have accumulated results in algorithm development across a wide range of fields, allowing us to leverage this potential to undertake problem-solving algorithm development projects. Furthermore, we can also accept contracts for chip implementation.
