Image and text data annotation for an autonomous driving recognition model
Participated in an image and text data annotation for a Chinese car brand, processing over 50,000 samples in total. • Formulated and implemented a unified annotation standard, and after cross-validation by the team, the accuracy rate was increased to over 95%. • Completed the annotation of 10,000 pieces of data within two months, providing a high-quality dataset for the training of the autonomous driving recognition model. • Through iterative feedback, the model's recognition error rate was reduced from 30% to 8%, significantly improving the training effect.