Sorry, you need to enable JavaScript to visit this website.

This document contains supplementary material which outlines key aspects regarding the dataset creation process, along with the methodology employed for benchmarking state-of-the-art deep leaning models. We aim to provide a clear and comprehensive overview, thus dividing the topic into six fundamental points. Beginning with point one, we delve into the method of data collection and organization, elucidating how the data can be interpreted and utilized. This section is crucial for understanding the origin and nature of the data comprising our dataset.

appendix

Accurate 3D lane detection from monocular images is crucial for autonomous driving. Recent advances leverage either front-view (FV) or bird’s-eye-view (BEV) features for prediction, inevitably limiting their ability to perceive driving environments precisely and resulting in suboptimal performance. To overcome the limitations of using features from a single view, we design a novel dual-view cross-attention mechanism, which leverages features from FV and BEV simultaneously. Based on this mechanism, we propose 3DLaneFormer, a powerful framework for 3D lane detection.

Multi-class multi-instance segmentation is the task of identifying masks for multiple object classes and multiple instances of the same class within an image. The Segment Anything Model (SAM) is a new foundation model designed for promptable multi-class multi-instance segmentation. SAM is able to segment objects in any image using a pre-defined point grid as an input prompt in the ``everything'' mode. However, out of the box SAM tends to output part or sub-part segmentation masks (under-segmentation) in different real-world applications.

In this work, we present SynTable, a Python-based dataset generator built using NVIDIA's Isaac Sim Replicator Composer for generating high-quality synthetic datasets for unseen object amodal instance segmentation of cluttered tabletop scenes. Our tool renders complex 3D scenes containing object meshes, materials, textures, lighting, and backgrounds. Metadata, including modal and amodal instance segmentation masks, occlusion masks, depth maps, and bounding boxes can be automatically generated based on user requirements.

We address distinguishing whether an input is a facial image by learning only a facial-expression recognition (FER) dataset.

We address distinguishing whether an input is a facial image by learning only a facial-expression recognition (FER) dataset.

With the advent of Deep Learning, people have begun to use it with computer vision approaches to identify plant diseases on a large scale targeting multiple crops and diseases. However, this requires a large amount of plant disease data, which is often not readily available, and the cost of acquiring disease images is high. Thus, developing a generalized model for recognizing unseen classes is very important and remains a major challenge to date. Existing methods solve the problem with general supervised recognition tasks based on the seen composition of the crop and the disease.

Capturing subtle visual differences between subordinate categories is crucial for improving the performance of Finegrained Visual Classification (FGVC). Recent works proposed deep learning models based on Vision Transformer (ViT) to take advantage of its self-attention mechanism to locate important regions of the objects and extract global information. However, their large number of layers with self-attention mechanism requires intensive computational cost and makes them impractical to be deployed on resource-restricted hardware including internet of things (IoT) devices.

In recent years, automotive radar has become an integral part of the advanced safety sensor stack. Although radar gives a significant advantage over a camera or Lidar, it suffers from poor angular resolution, unwanted noises and significant object smearing across the angular bins, making radar-based object detection challenging. We propose a novel radar-based object detection utilizing a deep learning-based super-resolution (DLSR) model. Due to the unavailability of low-high resolution radar data pair, we first simulate the data to train a DLSR model.