IQScan: Automatic Insect Loading and 3D Scanning.

We aim to tackle the challenges of digitising natural collections, especially insect collections, to extract rich information for bio-discovery and biosecurity. Although significant progress has been made in image-based 3D modelling of the specimens, particularly recent developments in the neural radiance field, capturing high-quality images at scale remains a major bottleneck for very large insect collections which can hold over tens of millions of specimens. Recent development in 3D insect scanning suggests that 3D printing allows custom 3D scanners to be built with lower cost and better functionalities. Current designs still require significant time-consuming manual work to operate the scanners. Furthermore, some specimens bend or slide with the change of gravity direction during scanning, causing errors in 3D modelling. This paper presents a new 3D design for insect specimen scanning along with a low-cost 3Dprinted prototype called IQScan, that allows a batch of multiple specimens to be loaded and scanned, to automate and speed up the image-capturing process. Major improvements include an autoloading mechanism to load multiple insects at a time. In addition, during scanning, tilting motion is applied to the camera, instead of the specimen, to keep the specimen undisturbed, which is especially important for flexible specimens. Early experiments show that IQScan can work with a batch of six insect specimens at a time and capture high-quality images for high-quality 3D insect models. IQSCan achieves a significant speedup of six times as compared to the existing scAnt 3D insect scanner. Further extension is possible for IQScan to work continuously with an unlimited number of insects. Our initial experiments show that the scanner works as expected producing usable images for 3D reconstructions.