ScratchDet: Training Single-Shot Object Detectors From Scratch

Abstract

Current state-of-the-art object objectors are fine-tuned from the off-the-shelf networks pretrained on large-scale classification dataset ImageNet, which incurs some additional problems: 1) The classification and detection have different degrees of sensitivity to translation, resulting in the learning objective bias; 2) The architecture is limited by the classification network, leading to the inconvenience of modification. To cope with these problems, training detectors from scratch is a feasible solution. However, the detectors trained from scratch generally perform worse than the pretrained ones, even suffer from the convergence issue in training. In this paper, we explore to train object detectors from scratch robustly. By analysing the previous work on optimization landscape, we find that one of the overlooked points in current trained-from-scratch detector is the BatchNorm. Resorting to the stable and predictable gradient brought by BatchNorm, detectors can be trained from scratch stably while keeping the favourable performance independent to the network architecture. Taking this advantage, we are able to explore various types of networks for object detection, without suffering from the poor convergence. By extensive experiments and analyses on downsampling factor, we propose the Root-ResNet backbone network, which makes full use of the information from original images. Our ScratchDet achieves the state-of-the-art accuracy on PASCAL VOC 2007, 2012 and MS COCO among all the train-from-scratch detectors and even performs better than several one-stage pretrained methods. Codes will be made publicly available at https: //github.com

CenterNet: Keypoint Triplets for Object Detection

“…We believe that our idea of adding an extra branch for the center keypoint can be potentially generalized to other existing one-stage approaches (e.g., SSD [27]). Meanwhile, some advanced training strategies [46] can be used for better performance. We leave as our future work.…”

Understanding the Effects of Pre-Training for Object Detectors via Eigenspectrum

“…(iii) The strides of conv5 x are too coarse to localize objects. DetNet [36] and ScratchDet [91] also discuss this problem and change the strides for object detection. Unlike these works, our finding is that SGD (with other regularization methods) automatically limits the intrinsic dimensionalities of standard ResNet without changing the strides.…”

Recent advances in deep learning for object detection

“…The generator tries to model data distribution by generating fake images using a noise vector input and use these fake images to confuse the discriminator, while the discriminator competes with the generator to identify the real images from fake images. GAN and its variants [155,156,157] There are some works [107,160,161] exploring training object detectors from scratch. Shen et al [107] first proposed a novel framework DSOD (Deeply Supervised Object Detectors) to train detectors from scratch.…”