search
HomeTechnology peripheralsAIUniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Written in front&Personal understanding

In recent years, vision-centered 3D perception in autonomous driving technology has developed rapidly. Although 3D perception models are structurally and conceptually similar, there are still gaps in feature representation, data formats, and objectives, which poses a challenge to design a unified and efficient 3D perception framework. Therefore, researchers need to work hard to address these gaps to achieve more accurate and reliable autonomous driving systems. Through collaboration and innovation, we hope to further improve the safety and performance of autonomous driving.

Especially for detection tasks and Occupancy tasks under BEV, it is very difficult to achieve joint training and achieve good results. This brings great troubles to many applications due to instability and effects that are difficult to control. However, UniVision is a simple and efficient framework that unifies the two main tasks of vision-centric 3D perception, namely occupancy prediction and object detection. The core of the framework is an explicit-implicit view transformation module for complementary 2D-3D feature transformation. In addition, UniVision also proposes a local global feature extraction and fusion module for efficient and adaptive voxel and BEV feature extraction, enhancement, and interaction. By adopting these methods, UniVision is able to achieve satisfactory results in detection tasks and Occupancy tasks under BEV.

UniVision proposes a joint occupancy detection data enhancement strategy and a progressive loss weight adjustment strategy to improve the efficiency and stability of multi-task framework training. Extensive experiments are conducted on four public benchmarks, including scene-free lidar segmentation, scene-free detection, OpenOccupancy and Occ3D. Experimental results show that UniVision achieved gains of 1.5 mIoU, 1.8 NDS, 1.5 mIoU and 1.8 mIoU on each benchmark respectively, reaching the SOTA level. Therefore, the UniVision framework can serve as a high-performance baseline for unified vision-centric 3D perception tasks.

The current state of the field of 3D perception

3D perception is the primary task of autonomous driving systems, which aims to utilize a series of sensors (such as lidar, radar and cameras) The data obtained can be used to comprehensively understand the driving scene and be used for subsequent planning and decision-making. In the past, the field of 3D perception has been dominated by lidar-based models due to the precise 3D information derived from point cloud data. However, lidar-based systems are costly, susceptible to severe weather, and inconvenient to deploy. In contrast, vision-based systems have many advantages, such as low cost, easy deployment, and good scalability. Therefore, vision-centered three-dimensional perception has attracted widespread attention from researchers.

Recently, vision-based 3D detection has made significant progress through improved feature representation transformation, temporal fusion, and supervision signal design, and the gap with LiDAR-based models continues to narrow. In addition, vision-based occupancy tasks have also developed rapidly in recent years. Unlike using 3D boxes to represent objects, occupancy can describe the geometric and semantic characteristics of the driving scene more comprehensively and is not limited by object shape and category.

Although detection methods and occupancy methods share structural and conceptual similarities, there is insufficient research on handling these two tasks simultaneously and exploring their interrelationships. Occupancy models and detection models usually extract different feature representations. The occupancy prediction task requires exhaustive semantic and geometric judgments, so voxel representations are widely used to preserve fine-grained 3D information. However, in detection tasks, BEV representation is more preferable since most objects lie on the same horizontal plane with smaller overlap.

Compared with BEV representation, voxel representation is higher in fineness, but less efficient. In addition, many advanced operators are mainly designed and optimized for 2D features, making their integration with 3D voxel representation not so simple. BEV representation is more advantageous in terms of time efficiency and memory efficiency, but it is suboptimal for dense spatial prediction because structural information is lost in the height dimension. In addition to feature representation, different perception tasks also differ in data formats and goals. Therefore, ensuring the uniformity and efficiency of training multi-task 3D perception frameworks is a huge challenge.

UniVision Network Structure

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

The overall architecture of the UniVision framework is shown in Figure 1. The framework receives multi-view images from surrounding N cameras as input and extracts image features through an image feature extraction network. Next, the Ex-Im view transformation module is used to convert 2D image features into 3D voxel features. This module combines depth-guided explicit feature boosting and query-guided implicit feature sampling. After view transformation, the voxel features are fed into the local global feature extraction and fusion block to extract local context-aware voxel features and global context-aware BEV features respectively. Next, information is exchanged on voxel features and BEV features for different downstream perception tasks through the cross-representation feature interaction module. During the training process, the UniVision framework uses combined Occ-Det data enhancement and progressive loss weight adjustment strategies for effective training. These strategies can improve the training effect and generalization ability of the framework. In short, the UniVision framework realizes the task of sensing the surrounding environment through the processing of multi-view images and 3D voxel features, as well as the application of feature interaction modules. At the same time, through the application of data enhancement and loss weight adjustment strategies, the training effect of the framework is effectively improved.

1) Ex-Im View Transform

Depth-oriented explicit feature enhancement. The LSS approach is followed here:

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

#2) Query-guided implicit feature sampling. However, there are some drawbacks in representing 3D information. The accuracy of is highly correlated with the accuracy of the estimated depth distribution. Furthermore, the points generated by LSS are not evenly distributed. Points are densely packed near the camera and sparse at distance. Therefore, we further use query-guided feature sampling to compensate for the above shortcomings.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Compared to points generated from LSS, voxel queries are uniformly distributed in 3D space, and they are learned from the statistical properties of all training samples, which is consistent with The depth prior information used in LSS is irrelevant. Therefore, and complement each other, connect them as the output features of the view transformation module:

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

2) Local and global feature extraction and fusion

Given input voxel features, first overlay the features on the Z-axis and use convolutional layers to reduce channels to obtain BEV features:

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Then, The model is divided into two parallel branches for feature extraction and enhancement. Local feature extraction, global feature extraction, and the final cross-representation feature interaction! As shown in Figure 1(b).

3) Loss function and detection head

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Progressive loss weight adjustment strategy. In practice, it is found that directly incorporating the above losses often causes the training process to fail and the network to fail to converge. In the early stages of training, voxel features Fvoxel are randomly distributed, and supervision in the occupancy head and detection head contributes less than other losses in convergence. At the same time, loss items such as the classification loss Lcls in the detection task are very large and dominate the training process, making it difficult to optimize the model. To overcome this problem, a progressive loss weight adjustment strategy is proposed to dynamically adjust the loss weight. Specifically, the control parameter δ is added to the non-image-level losses (i.e., occupancy loss and detection loss) to adjust the loss weight in different training epochs. The control weight δ is set to a small value Vmin at the beginning and gradually increases to Vmax over N training epochs:

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

4) Combined with Occ-Det spatial data enhancement

In 3D detection tasks, in addition to common image-level data enhancement, spatial-level data enhancement is also effective in improving model performance. Effective. However, applying spatial level enhancement in occupancy tasks is not straightforward. When we apply data augmentation (such as random scaling and rotation) to discrete occupancy labels, it is difficult to determine the resulting voxel semantics. Therefore, existing methods only apply simple spatial augmentation such as random flipping in occupancy tasks.

To solve this problem, UniVision proposes a joint Occ-Det spatial data enhancement to allow simultaneous enhancement of 3D detection tasks and occupancy tasks in the framework. Since the 3D box labels are continuous values ​​and the enhanced 3D box can be directly calculated for training, the enhancement method in BEVDet is followed for detection. Although occupancy labels are discrete and difficult to manipulate, voxel features can be treated as continuous and can be processed through operations such as sampling and interpolation. It is therefore recommended to transform voxel features instead of directly operating on occupancy labels for data augmentation.

Specifically, spatial data augmentation is first sampled and the corresponding 3D transformation matrix is ​​calculated. For the occupancy labels and their voxel indices , we calculate their three-dimensional coordinates. Then, apply and normalize it to obtain the voxel indices in the enhanced voxel feature :

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Experiment Comparison of results

Used multiple data sets for verification, NuScenes LiDAR Segmentation, NuScenes 3D Object Detection, OpenOccupancy and Occ3D.

NuScenes LiDAR Segmentation: According to the recent OccFormer and TPVFormer, camera images are used as input for the lidar segmentation task, and the lidar data is only used to provide 3D locations for querying the output features. Use mIoU as the evaluation metric.

NuScenes 3D Object Detection: For detection tasks, use the official metric of nuScenes, the nuScene Detection Score (NDS), which is the weighted sum of average mAP and several metrics, including average translation error (ATE), average Scale Error (ASE), Average Orientation Error (AOE), Average Velocity Error (AVE) and Average Attribute Error (AAE).

OpenOccupancy: The OpenOccupancy benchmark is based on the nuScenes dataset and provides semantic occupancy labels at 512×512×40 resolution. The labeled classes are the same as those in the lidar segmentation task, using mIoU as the evaluation metric!

Occ3D: The Occ3D benchmark is based on the nuScenes dataset and provides semantic occupancy labels at 200×200×16 resolution. Occ3D further provides visible masks for training and evaluation. The labeled classes are the same as those in the lidar segmentation task, using mIoU as the evaluation metric!

1) Nuscenes LiDAR segmentation

Table 1 shows the results of the nuScenes LiDAR segmentation benchmark. UniVision significantly outperforms the state-of-the-art vision-based method OccFormer by 1.5% mIoU and sets a new record for vision-based models on the leaderboard. Notably, UniVision also outperforms some lidar-based models such as PolarNe and DB-UNet.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

2) NuScenes 3D object detection task

As shown in Table 2, when using the same training settings for fair comparison , UniVision was shown to outperform other methods. Compared with BEVDepth at 512×1408 image resolution, UniVision achieves gains of 2.4% and 1.1% in mAP and NDS respectively. When the model is scaled up and UniVision is combined with temporal input, it further outperforms SOTA-based temporal detectors by significant margins. UniVision achieves this with a smaller input resolution, and it does not use CBGS.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

3) Comparison of OpenOccupancy results

The results of the OpenOccupancy benchmark test are shown in Table 3. UniVision significantly outperforms recent vision-based occupancy methods including MonoScene, TPVFormer, and C-CONet in terms of mIoU by 7.3%, 6.5%, and 1.5%, respectively. Furthermore, UniVision outperforms some lidar-based methods such as LMSCNet and JS3C-Net.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

4) Occ3D experimental results

Table 4 lists the results of the Occ3D benchmark test. UniVision significantly outperforms recent vision-based methods in terms of mIoU under different input image resolutions, by more than 2.7% and 1.8% respectively. It is worth noting that BEVFormer and BEVDet-stereo load pre-trained weights and use temporal inputs in inference, while UniVision does not use them but still achieves better performance.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

#5) Effectiveness of components in detection tasks

Ablation studies for detection tasks are shown in Table 5. When the BEV-based global feature extraction branch is inserted into the baseline model, the performance improves by 1.7% mAP and 3.0% NDS. When the voxel-based occupancy task is added to the detector as an auxiliary task, the model’s mAP gain increases by 1.6%. When cross-representation interactions are explicitly introduced from voxel features, the model achieves the best performance, improving mAP and NDS by 3.5% and 4.2%, respectively, compared to the baseline;

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

6) Effectiveness of components in the occupancy task

Ablation studies for the occupancy task are shown in Table 6. The voxel-based local feature extraction network brings an improvement of 1.96% mIoU gain to the baseline model. When the detection task is introduced as an auxiliary supervision signal, the model performance improves by 0.4% mIoU.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

7) Others

Table 5 and Table 6 show that in the UniVision framework, detection tasks and occupancy tasks complement each other of. For detection tasks, occupancy supervision can improve mAP and mATE metrics, indicating that voxel semantic learning effectively improves the detector's perception of object geometry, i.e., centrality and scale. For the occupancy task, detection supervision significantly improves the performance of the foreground category (i.e., the detection category), resulting in an overall improvement.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

The effectiveness of the combined Occ-Det spatial enhancement, Ex-Im view conversion module and progressive loss weight adjustment strategy is shown in Table 7. With the proposed spatial augmentation and the proposed view transformation module, it shows significant improvements in detection tasks and occupancy tasks on mIoU, mAP and NDS metrics. The loss weight adjustment strategy can effectively train the multi-task framework. Without this, the training of the unified framework cannot converge and the performance is very low.

UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!

Original link: https://mp.weixin.qq.com/s/8jpS_I-wn1-svR3UlCF7KQ


The above is the detailed content of UniVision introduces a new generation of unified framework: BEV detection and Occupancy dual tasks reach the most advanced level!. For more information, please follow other related articles on the PHP Chinese website!

Statement
This article is reproduced at:51CTO.COM. If there is any infringement, please contact admin@php.cn delete
从VAE到扩散模型:一文解读以文生图新范式从VAE到扩散模型:一文解读以文生图新范式Apr 08, 2023 pm 08:41 PM

1 前言在发布DALL·E的15个月后,OpenAI在今年春天带了续作DALL·E 2,以其更加惊艳的效果和丰富的可玩性迅速占领了各大AI社区的头条。近年来,随着生成对抗网络(GAN)、变分自编码器(VAE)、扩散模型(Diffusion models)的出现,深度学习已向世人展现其强大的图像生成能力;加上GPT-3、BERT等NLP模型的成功,人类正逐步打破文本和图像的信息界限。在DALL·E 2中,只需输入简单的文本(prompt),它就可以生成多张1024*1024的高清图像。这些图像甚至

找不到中文语音预训练模型?中文版 Wav2vec 2.0和HuBERT来了找不到中文语音预训练模型?中文版 Wav2vec 2.0和HuBERT来了Apr 08, 2023 pm 06:21 PM

Wav2vec 2.0 [1],HuBERT [2] 和 WavLM [3] 等语音预训练模型,通过在多达上万小时的无标注语音数据(如 Libri-light )上的自监督学习,显著提升了自动语音识别(Automatic Speech Recognition, ASR),语音合成(Text-to-speech, TTS)和语音转换(Voice Conversation,VC)等语音下游任务的性能。然而这些模型都没有公开的中文版本,不便于应用在中文语音研究场景。 WenetSpeech [4] 是

普林斯顿陈丹琦:如何让「大模型」变小普林斯顿陈丹琦:如何让「大模型」变小Apr 08, 2023 pm 04:01 PM

“Making large models smaller”这是很多语言模型研究人员的学术追求,针对大模型昂贵的环境和训练成本,陈丹琦在智源大会青源学术年会上做了题为“Making large models smaller”的特邀报告。报告中重点提及了基于记忆增强的TRIME算法和基于粗细粒度联合剪枝和逐层蒸馏的CofiPruning算法。前者能够在不改变模型结构的基础上兼顾语言模型困惑度和检索速度方面的优势;而后者可以在保证下游任务准确度的同时实现更快的处理速度,具有更小的模型结构。陈丹琦 普

解锁CNN和Transformer正确结合方法,字节跳动提出有效的下一代视觉Transformer解锁CNN和Transformer正确结合方法,字节跳动提出有效的下一代视觉TransformerApr 09, 2023 pm 02:01 PM

由于复杂的注意力机制和模型设计,大多数现有的视觉 Transformer(ViT)在现实的工业部署场景中不能像卷积神经网络(CNN)那样高效地执行。这就带来了一个问题:视觉神经网络能否像 CNN 一样快速推断并像 ViT 一样强大?近期一些工作试图设计 CNN-Transformer 混合架构来解决这个问题,但这些工作的整体性能远不能令人满意。基于此,来自字节跳动的研究者提出了一种能在现实工业场景中有效部署的下一代视觉 Transformer——Next-ViT。从延迟 / 准确性权衡的角度看,

Stable Diffusion XL 现已推出—有什么新功能,你知道吗?Stable Diffusion XL 现已推出—有什么新功能,你知道吗?Apr 07, 2023 pm 11:21 PM

3月27号,Stability AI的创始人兼首席执行官Emad Mostaque在一条推文中宣布,Stable Diffusion XL 现已可用于公开测试。以下是一些事项:“XL”不是这个新的AI模型的官方名称。一旦发布稳定性AI公司的官方公告,名称将会更改。与先前版本相比,图像质量有所提高与先前版本相比,图像生成速度大大加快。示例图像让我们看看新旧AI模型在结果上的差异。Prompt: Luxury sports car with aerodynamic curves, shot in a

五年后AI所需算力超100万倍!十二家机构联合发表88页长文:「智能计算」是解药五年后AI所需算力超100万倍!十二家机构联合发表88页长文:「智能计算」是解药Apr 09, 2023 pm 07:01 PM

人工智能就是一个「拼财力」的行业,如果没有高性能计算设备,别说开发基础模型,就连微调模型都做不到。但如果只靠拼硬件,单靠当前计算性能的发展速度,迟早有一天无法满足日益膨胀的需求,所以还需要配套的软件来协调统筹计算能力,这时候就需要用到「智能计算」技术。最近,来自之江实验室、中国工程院、国防科技大学、浙江大学等多达十二个国内外研究机构共同发表了一篇论文,首次对智能计算领域进行了全面的调研,涵盖了理论基础、智能与计算的技术融合、重要应用、挑战和未来前景。论文链接:​https://spj.scien

​什么是Transformer机器学习模型?​什么是Transformer机器学习模型?Apr 08, 2023 pm 06:31 PM

译者 | 李睿审校 | 孙淑娟​近年来, Transformer 机器学习模型已经成为深度学习和深度神经网络技术进步的主要亮点之一。它主要用于自然语言处理中的高级应用。谷歌正在使用它来增强其搜索引擎结果。OpenAI 使用 Transformer 创建了著名的 GPT-2和 GPT-3模型。自从2017年首次亮相以来,Transformer 架构不断发展并扩展到多种不同的变体,从语言任务扩展到其他领域。它们已被用于时间序列预测。它们是 DeepMind 的蛋白质结构预测模型 AlphaFold

AI模型告诉你,为啥巴西最可能在今年夺冠!曾精准预测前两届冠军AI模型告诉你,为啥巴西最可能在今年夺冠!曾精准预测前两届冠军Apr 09, 2023 pm 01:51 PM

说起2010年南非世界杯的最大网红,一定非「章鱼保罗」莫属!这只位于德国海洋生物中心的神奇章鱼,不仅成功预测了德国队全部七场比赛的结果,还顺利地选出了最终的总冠军西班牙队。不幸的是,保罗已经永远地离开了我们,但它的「遗产」却在人们预测足球比赛结果的尝试中持续存在。在艾伦图灵研究所(The Alan Turing Institute),随着2022年卡塔尔世界杯的持续进行,三位研究员Nick Barlow、Jack Roberts和Ryan Chan决定用一种AI算法预测今年的冠军归属。预测模型图

See all articles

Hot AI Tools

Undresser.AI Undress

Undresser.AI Undress

AI-powered app for creating realistic nude photos

AI Clothes Remover

AI Clothes Remover

Online AI tool for removing clothes from photos.

Undress AI Tool

Undress AI Tool

Undress images for free

Clothoff.io

Clothoff.io

AI clothes remover

AI Hentai Generator

AI Hentai Generator

Generate AI Hentai for free.

Hot Tools

EditPlus Chinese cracked version

EditPlus Chinese cracked version

Small size, syntax highlighting, does not support code prompt function

VSCode Windows 64-bit Download

VSCode Windows 64-bit Download

A free and powerful IDE editor launched by Microsoft

ZendStudio 13.5.1 Mac

ZendStudio 13.5.1 Mac

Powerful PHP integrated development environment

MantisBT

MantisBT

Mantis is an easy-to-deploy web-based defect tracking tool designed to aid in product defect tracking. It requires PHP, MySQL and a web server. Check out our demo and hosting services.

SublimeText3 Chinese version

SublimeText3 Chinese version

Chinese version, very easy to use