Recent years have seen an increase in the use of gigapixel-level image and video capture systems and benchmarks with high-resolution wide (HRW) shots. However, unlike close-up shots in the MS COCO dataset, the higher resolution and wider field of view raise unique challenges, such as extreme sparsity and huge scale changes, causing existing close-up detectors inaccuracy and inefficiency. In this paper, we present a novel model-agnostic sparse vision transformer, dubbed SparseFormer, to bridge the gap of object detection between close-up and HRW shots. The proposed SparseFormer selectively uses attentive tokens to scrutinize the sparsely distributed windows that may contain objects. In this way, it can jointly explore global and local attention by fusing coarse- and fine-grained features to handle huge scale changes. SparseFormer also benefits from a novel Cross-slice non-maximum suppression (C-NMS) algorithm to precisely localize objects from noisy windows and a simple yet effective multi-scale strategy to improve accuracy. Extensive experiments on two HRW benchmarks, PANDA and DOTA-v1.0, demonstrate that the proposed SparseFormer significantly improves detection accuracy (up to 5.8%) and speed (up to 3x) over the state-of-the-art approaches.

CornerNet is a different object detection technique where we detects the objects bounding box by a paired key-points, the top-left corner and the bottom-right corner using a single convolution neural network. By detecting the key points, it eliminates the need of different anchor boxes commonly used in single stage detectors. In this paper by Hei Law and Jia Deng from Princeton University, they have introduced a new approach to object detection which outperforms all the single stage detectors. CornetNet introduces a new type of pooling layer called Corner Pooling, that helps localizing the corners. The Net achieves 42.2% AP on MS COCO dataset.

A system is developed to warn the driver of unrecognised vehicles in the surrounding vehicles. A system is developed to warn the driver of unrecognised vehicles in the surrounding vehicles.

Object Detection in a video is to locate the presence of objects, put them under certain classes based on our deep learning model, and place bounding boxes around them. To simply put our Input is a video or images with multiple objects and Output is the same video or image with bounding boxes (of certain height and width) around the objects class names and probabilities to which they belong. Here we will be using a pre-trained YOLO (You Only Look Once) model which was trained for a large dataset of around 80 classes of objects for a long time with high-powered resources. This article, will not go much deeper into the YOLO architecture and focus more on the ImageAI library to do object detection on our video and get the results. Building these models from scratch takes a lot of understanding of mathematics and its architecture along with thousands of lines of code.

We present a scalable approach for Detecting Objects by transferring Common-sense Knowledge (DOCK) from source to target categories. In our setting, the training data for the source categories have bounding box annotations, while those for the target categories only have image-level annotations. Current state-of-the-art approaches focus on image-level visual or semantic similarity to adapt a detector trained on the source categories to the new target categories. In contrast, our key idea is to (i) use similarity not at the image-level, but rather at the region-level, and (ii) leverage richer common-sense (based on attribute, spatial, etc.) to guide the algorithm towards learning the correct detections. We acquire such common-sense cues automatically from readily-available knowledge bases without any extra human effort. On the challenging MS COCO dataset, we find that common-sense knowledge can substantially improve detection performance over existing transfer-learning baselines.