The Conference on Computer Vision and Pattern Recognition (CVPR) is among the three top meetings (CVPR, ICCV, and ECCV) in the computer vision field. The past five years have witnessed the continued growth of CVPR's paper submissions. According to CVPR's official website, more than 7,500 papers were received in 2021, with 1,663 accepted, an acceptance rate of 27.3 percent.
The paper of AInnovation accepted by CVPR in 2021 proposes zero-sample instance segmentation, which is a natural extension of zero-sample object detection (similar to Mask-RCNN to Faster-RCNN).
In the process of AI commercialization, AInnovation will encounter very real problems of AI implementation. Especially, when facing manufacturing scenarios with limited informatization, complex and diverse samples, a critical shortage of positive samples, or a need for very specialized labeling methods,
many AI companies will be shackled by industrial challenges such as difficult data acquisition and high cost of labeling in their pursuit for fast commercialization.
After intensive research, AInnovation has made groundbreaking, innovative results, which will greatly promote data discovery, data coarse screening, auxiliary labeling, the exploration of basic model capabilities, etc.