LaSagnA: Language-based Segmentation Assistant for Complex Queries
This work addresses the problem of complex query processing in vision-language models for researchers and practitioners in computer vision, representing an incremental improvement with novel training strategies.
The paper tackles the limitations of vision-language models in handling multiple targets per query and detecting absent objects by introducing a general sequence format for complex queries and incorporating semantic segmentation into the training pipeline, achieving comparable results to conventional methods on segmentation datasets and outperforming other vision-language models in reasoning and referring segmentation tasks.
Recent advancements have empowered Large Language Models for Vision (vLLMs) to generate detailed perceptual outcomes, including bounding boxes and masks. Nonetheless, there are two constraints that restrict the further application of these vLLMs: the incapability of handling multiple targets per query and the failure to identify the absence of query objects in the image. In this study, we acknowledge that the main cause of these problems is the insufficient complexity of training queries. Consequently, we define the general sequence format for complex queries. Then we incorporate a semantic segmentation task in the current pipeline to fulfill the requirements of training data. Furthermore, we present three novel strategies to effectively handle the challenges arising from the direct integration of the proposed format. The effectiveness of our model in processing complex queries is validated by the comparable results with conventional methods on both close-set and open-set semantic segmentation datasets. Additionally, we outperform a series of vLLMs in reasoning and referring segmentation, showcasing our model's remarkable capabilities. We release the code at https://github.com/congvvc/LaSagnA.