Image is progressively enhanced through multiple GDIP blocks, each guided by a different layer Multi-level GDIP-Yolo (MGDIP-Yolo): a multi-level version of GDIP where an.GDIP-Yolo: a single GDIP block (fed with an adverse input and latent embeddings from an encoder)Ĭan be plugged into existing object detection networks (e.g., Yolo) and trained end-to-end withĪdverse condition images (fog and low-lighting).Using our GDIP block, we propose novel variants as below: ThisĮliminates any image enhancement overhead during inference, unlike other state-of-the-art works, thus resulting in higher throughput. In addition, GDIP block has a unique advantage with its utility as a training regularizer, which directly improves object detection training for adverse conditions.GDIP block can be integrated as single or multi-level enhancement module with an encoder.GDIP provides flexible concurrent gated weighting of the individual IP operations that result in superior detection performance.We present Gated Differentiable Image Processing (GDIP) block, a domain-agnostic architecture which adaptively enhances adverse images for object detection by performing concurrent gated-weighing of image.Real-world foggy (RTTS dataset) and dark (ExDark dataset) images with annotations ThisĮnhanced output is sent to the downstream object detection network, leading to superior detection performance We tackle this problem by proposing a Gated Differentiable Image Processing (GDIP) block, whichĮnhances the image by performing weighted Image Processing (IP) operations concurrently on the adverse input image. Object detection is challenging under adverse weather conditions (such as fog or low-lighting)īecause the objects are partially or not visible, resulting in missed detections. GDIP: Object-Detection in Adverse Weather Conditions Using Gated Differentiable Image Processingġ Robotics Research Centre, IIIT Hyderabad, India.Ģ Australian Centre for Robotic Vision at the Queensland University of Technology (QUT), Brisbane, Australia.ģ ZF Friedrichshafen TCI, Hyderabad, India
0 kommentar(er)
