Search

Traditional closed-world object detection methods are limited to a predefined set of classes and struggle to recognize objects beyond these boundaries. This work proposes an improved Open-World Object Detection (OWOD) methodology that supports the identification and Incremental Detection of both known and unknown objects. Existing OWOD methods typically face uncertainties due to the adoption of complex and ambiguous strategies for discovering the unknown objects. Such approaches lower the detection accuracy of the known classes. To effectively tackle these challenges, this study presents a new Open-World Novelty and Outlier Detection (OWNOD). The framework integrates a dynamic thresholding mechanism a customized loss function and statistical verification technique optimize the simultaneous detection of both known and unknown objects. By combining these approaches, OWNOD enhances the balance between identifying new, unseen objects and maintaining accuracy for previously learned classes, effectively addressing the inherent complexities of open-world object detection. We benchmark against state-of-the-art methods, that includes the baseline ORE and ABLTL. We also report significant improvements across multiple metrics.For instance, in Task 1, K-mAP improved by 3.10 points, U-mAP by 2.20 points, and U-Recall by 9.50 points. In Task 2, K-mAP improved by 5.65 points, U-mAP by 2.43 points, and U-Recall by 2.46 points. In Task 3, we achieved gains of 4.14 points for K-mAP, 0.34 points for U-mAP, and 6.62 points for U-Recall. For Task 4, where all classes are known, we report a K-mAP improvement of 5.89 points, demonstrating the robustness and effectiveness of our model.

Open-world object detection is a challenging set of tasks in the realm of computer vision. In these tasks, the object detection model processes the input image or video and undergoes inference to identify objects or features present in the image or video. The main objective of the model is to identify the seen and unseen classes rather than identifying only the classes that that were introduced to it during training. The challenging task is to create instance-level true bounding boxes tightly around the true objects and classify and localize them with their true class labels, without missing any of the true object or assigning false positive labels to them. Motivated by the pressing need to advance the capabilities of open-world object detection we present a novel clustering technique called margin-based latent space clustering and deployed it in the classification head of the Faster RCNN. Furthermore, we propose a novel loss function called margin-based loss coupled with regularization parameters aiming to optimize the outlier identification. The proposed method is also capable of incremental learning. Overall, all four incremental tasks are assessed by employing benchmark evaluation metrics. The proposed method outperforms the existing state-of-the-art method, with significant improvements in mean average precision (mAP). We improved the mAP by 4.49% on task 1, 3.71% on task 2, 6.19% on task 3 and 2.81% on task 4. We also reduced the ‘Wilderness Impact and Absolute Open-Set Error’ metrics (For assessing the false positive detection, both metrics are employed), on every task.