Computer Vision Data Annotation QA – Fish Detection Dataset

We are looking for one detail-oriented data annotator to perform quality control (QA) on a pre-annotated computer vision dataset containing fish images. The dataset already contains AI-generated point annotations. The task is not to annotate everything from scratch. Instead, the selected annotator must carefully review each image and verify that every visible fish is correctly marked. Main tasks: Review existing point annotations on each image. Accept correctly annotated fish. Remove false-positive annotations where no fish is present. Add missing annotations when a fish has not been detected. Ensure that each visible fish has exactly one annotation. Maintain high consistency and accuracy throughout the complete dataset. Dataset: Approximately 2,300 images in total. Approximately 80,000–90,000 existing fish annotations across the complete dataset. Images are already pre-annotated by an AI model. The selected freelancer will be responsible for reviewing the entire dataset. This is primarily a visual verification and data annotation QA task. No programming, machine-learning model development, or model training is required. Requirements: Previous experience with image annotation, data labeling, computer vision datasets, or similar QA work is preferred. Strong attention to detail. Ability to work consistently on repetitive visual tasks. Good communication and reliability. Ability to work several hours per day over the next days. Ability to complete the full dataset within approximately 7–10 days, while maintaining a high level of accuracy and consistency. Paid trial: The selected candidate may first receive a small paid trial batch of approximately 50–100 images. The trial will be used to evaluate: Accuracy. Ability to identify false positives. Ability to identify missing fish annotations. Consistency. Attention to detail. Overall quality of the annotation review. If the quality of the trial meets our requirements, the freelancer will continue with the complete dataset. We will provide: Access to the annotation interface. Clear instructions and examples. The complete set of images to be reviewed. Existing AI-generated annotations. Support for questions or ambiguous cases. Example files attached: 01_fish_annotation_QA_interface.png 02_fish_point_annotation_detail.png When applying, please include: Your previous experience with image annotation, data labeling, or similar work. Tools you have used, such as CVAT, Roboflow, Label Studio, or similar, if applicable. Your availability over the next 7–10 days. How many hours per day you can dedicate to this project. Your estimated turnaround time for the complete dataset of approximately 2,300 images. Confirmation that you are comfortable performing repetitive visual QA work with a strong focus on accuracy.