Deep learning is a sub-field of Machine Learning and a significant branch of Artificial Intelligence. It aims to infer high-level abstractions from raw data by using a deep graph with multiple processing layers composed of multiple linear and non-linear transformations. Deep learning is loosely based on models of information processing and communication in the brain. Deep learning replaces handcrafted features with ones learned from very large amounts of annotated data. Learning occurs by iteratively estimating hundreds of thousands of parameters in the deep graph with efficient algorithms.

Several deep learning architectures such as convolutional deep neural networks (CNNs), and recurrent neural networks have been applied to computer vision, speech recognition, natural language processing, and audio recognition to produce state-of-the-art results on various tasks.

ProtoTech is one of the pioneering companies to apply deep learning principles to 3D data. 3D Measure Up service uses deep learning in very intuitive ways to process the input body data and derive meaningful information viz. Landmarks, measurements etc. .

No. With 3D Measure Up, you don’t have to build, maintain or upgrade deep learning pipelines.

To achieve accurate results on complex 3D scans such as feature detection, mesh processing, point cloud manipulations, geometry computations, deep learning systems need to be tuned properly. 3D Measure Up is fully managed and comes pre-trained for 3D body scan recognition and measurement tasks so that you don’t have to invest your time and resources on creating a deep learning pipeline. 3D Measure Up continues to improve the accuracy of its models by building upon the latest research and sourcing new training data. Please note that we do not use your data for training our algorithms.

There are a number of ways this can be done. Following is one of the ways:
1. Download and install Meshlab
2. Import 3D body scan file. => The number of polygons it contains is shown at the bottom of the screen.
3. Go to “Filters” > “Remeshing, simplification and construction” > “Quadratic Edge Collapse Detection”=>A Dialog pops up with options to adjust the mesh.

The options are :
A. Target number of faces: Here type the number of polygons you wish your file to have. This procedure is called reduction of polygons also known as mesh decimation.
B. Quality threshold: You can select a number between 0 and 1. The recommended entry is 1 as it gives satisfying results. This parameter affects the original model’s shape (ex. bad shaped faces etc. ). So in order for the software to calculate the original shape of the model with only well shaped faces, it requires to enter a high number of faces as it allows more freedom in the final triangle shape.
C. Preserve Boundary of the Mesh: Here you should click on this option, as it has to do with preserving the mesh boundaries. Since your file is going under a simplification process, it is essential not to destroy boundaries of the mesh, such as exposed edges etc..
D. Preserve Normal: Here you should also select this option if it is not already on by default. With this action you prevent the software to flip face normals and preserves the orientation of the surface. Optimal position of simplified vertices: This option should also be on by default. It is essential for preventing the edges from collapsing and substituting the original mesh. Planar simplification: This option should be activated when there are flat surfaces on the model that you want to be tessellated. This parameter affects the quality of the shape of the final triangles on planar portions of the mesh and improves the accuracy/complexity ratio.

• The scan model should be a single mesh in STL, OBJ, PLY or GLB file format and less than 50MB in size (we recommend between 10MB to 20MB for generally accepted precision).
• The model should be either a full-body scan or supported partial-body scans only, and a scan of a single person. Multiple objects or persons in a scan will not yield proper results.
• The model should be aligned with feet flat on the ground, height along the +Y axis, and face along the +Z axis for better accuracy of measurements. If the model is not aligned, please use the automatic alignment feature before requesting measurements.
• The mesh should be watertight with no holes, gaps, or missing faces, clean without any stray triangles or self-intersecting surfaces, and made of triangles and not quadrilaterals/polygons.
• We recommend an upright ‘A’ pose for the model’s pose. Other postures are supported, but we may not be able to extract all the landmarks or it may impact the accuracy.

To acquire a 3D scan of the person/user, you can use a LIDAR-based scanning app or apps using libraries like Meshcapade. Once you have the scan, you can upload it to any HTTP server or cloud storage such as Google Drive or Dropbox file URL. The URL of the scan is passed as a parameter to our API, which retrieves the scan for generating measurements. The landmark details and measurements are output as a JSON stream for you to use in the way you wish. Refer to 3D Measure Up API

3D Measure Up provides a dashboard for monitoring token usage and subscription status, allowing better planning and utilization. By logging in via desktop browser, users can view token usage in the top right corner of the application and click for more details, including available tokens and active subscriptions. You can find detailed, easy-to-follow instructions on Our Blog.