LatteCLIP: Unsupervised CLIP Fine-Tuning via LMM-Synthetic Texts

Large-scale vision-language pre-trained (VLP) models (e.g., CLIP) are renowned for their versatility, as they can be applied to diverse applications in a zero-shot setup. However, when these models are used in specific domains, their performance often falls short due to domain gaps or the under-representation of these domains in the training data. While fine-tuning VLP models on custom datasets with human-annotated labels can address this issue, annotating even a small-scale dataset (e.g., 100k samples) can be an expensive endeavor, often requiring expert annotators if the task is complex. To address these challenges, we propose LatteCLIP, an unsupervised method for fine-tuning CLIP models on classification with known class names in custom domains, without relying on human annotations. Our method leverages Large Multimodal Models (LMMs) to generate expressive textual descriptions for both individual images and groups of images. These provide additional contextual information to guide the fine-tuning process in the custom domains. Since LMM-generated descriptions are prone to hallucination or missing details, we introduce a novel strategy to distill only the useful information and stabilize the training. Specifically, we learn rich per-class prototype representations from noisy generated texts and dual pseudo-labels. Our experiments on 10 domain-specific datasets show that LatteCLIP outperforms pre-trained zero-shot methods by an average improvement of +4.74 points in top-1 accuracy and other state-of-the-art unsupervised methods by +3.45 points.

Cross-modal Learning for Domain Adaptation in 3D Semantic Segmentation

Domain adaptation is an important task to enable learning when labels are scarce. While most works focus only on the image modality, there are many important multi-modal datasets. In order to leverage multi-modality for domain adaptation, we propose cross-modal learning, where we enforce consistency between the predictions of two modalities via mutual mimicking. We constrain our network to make correct predictions on labeled data and consistent predictions across modalities on unlabeled target-domain data. Experiments in unsupervised and semi-supervised domain adaptation settings prove the effectiveness of this novel domain adaptation strategy. Specifically, we evaluate on the task of 3D semantic segmentation using the image and point cloud modality. We leverage recent autonomous driving datasets to produce a wide variety of domain adaptation scenarios including changes in scene layout, lighting, sensor setup and weather, as well as the synthetic-to-real setup. Our method significantly improves over previous uni-modal adaptation baselines on all adaption scenarios. Code will be made available.

xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation

Unsupervised Domain Adaptation (UDA) is crucial to tackle the lack of annotations in a new domain. There are many multi-modal datasets, but most UDA approaches are uni-modal. In this work, we explore how to learn from multi-modality and propose cross-modal UDA (xMUDA) where we assume the presence of 2D images and 3D point clouds for 3D semantic segmentation. This is challenging as the two input spaces are heterogeneous and can be impacted differently by domain shift. In xMUDA, modalities learn from each other through mutual mimicking, disentangled from the segmentation objective, to prevent the stronger modality from adopting false predictions from the weaker one. We evaluate on new UDA scenarios including day-to-night, country-to-country and dataset-to-dataset, leveraging recent autonomous driving datasets. xMUDA brings large improvements over uni-modal UDA on all tested scenarios, and is complementary to state-of-the-art UDA techniques.

Multi-view PointNet for 3D Scene Understanding

Fusion of 2D images and 3D point clouds is important because information from dense images can enhance sparse point clouds. However, fusion is challenging because 2D and 3D data live in different spaces. In this work, we propose MVPNet (Multi-View PointNet), where we aggregate 2D multi-view image features into 3D point clouds, and then use a point based network to fuse the features in 3D canonical space to predict 3D semantic labels. To this end, we introduce view selection along with a 2D-3D feature aggregation module. Extensive experiments show the benefit of leveraging features from dense images and reveal superior robustness to varying point cloud density compared to 3D-only methods. On the ScanNetV2 benchmark, our MVPNet significantly outperforms prior point cloud based approaches on the task of 3D Semantic Segmentation. It is much faster to train than the large networks of the sparse voxel approach. We provide solid ablation studies to ease the future design of 2D-3D fusion methods and their extension to other tasks, as we showcase for 3D instance segmentation.

End-to-End Race Driving with Deep Reinforcement Learning

We present research using the latest reinforcement learning algorithm for end-to-end driving without any mediated perception (object recognition, scene understanding). The newly proposed reward and learning strategies lead together to faster convergence and more robust driving using only RGB image from a forward facing camera. An Asynchronous Actor Critic (A3C) framework is used to learn the car control in a physically and graphically realistic rally game, with the agents evolving simultaneously on tracks with a variety of road structures (turns, hills), graphics (seasons, location) and physics (road adherence). A thorough evaluation is conducted and generalization is proven on unseen tracks and using legal speed limits. Open loop tests on real sequences of images show some domain adaption capability of our method.

Sparse and Dense Data with CNNs: Depth Completion and Semantic Segmentation

Convolutional neural networks are designed for dense data, but vision data is often sparse (stereo depth, point clouds, pen stroke, etc.). We present a method to handle sparse depth data with optional dense RGB, and accomplish depth completion and semantic segmentation changing only the last layer. Our proposal efficiently learns sparse features without the need of an additional validity mask. We show how to ensure network robustness to varying input sparsities. Our method even works with densities as low as 0.8% (8 layer lidar), and outperforms all published state-of-the-art on the Kitti depth completion benchmark.