bdd100k

MobileNets-SSD/SSDLite on VOC/BDD100K Datasets

Image2Image Translation Research

My Computer Vision project from my Computer Vision Course (Fall 2020) at Goethe University Frankfurt, Germany. Performance comparison between state-of-the-art Object Detection algorithms YOLO and Faster R-CNN based on the Berkeley DeepDrive (BDD100K) Dataset.

[WACV 2025] Official implementation of "RAW-Diffusion: RGB-Guided Diffusion Models for High-Fidelity RAW Image Generation"

Open source training framework for vision tasks. Scales up on data and scales up on tasks. Official Implementation for https://arxiv.org/abs/2310.00920

Faster R-CNN with KITTI, BDD100k support in PyTorch 1.0

This repository contains the implementation of a lane detection system using the UNet architecture. The model is trained on the BDD100K dataset, leveraging its diverse and large-scale data to ensure robust performance under various weather conditions and different times of day.

Pytorch implementation of DeepLab V3+

Road Object Detection using Deep Learning, based on tensorflow framework and BDD100k dataset

Perform inference with TwinLiteNet model using ONNX Runtime. TwinLiteNet is a lightweight and efficient deep learning model designed for drivable area and lane segmentation

Training FCOS on KITTI and BDD100K datasets for real-time traffic object detection with PyTorch.

University of Bristol MEng Computer Science Dissertation and Code for 'Predicting Ego-Vehicle Speed from Monocular Dash-Cam Video in Diverse Conditions'.

Code for paper: "Road object detection: a comparative study of deep learning-based algorithms" https://link.springer.com/article/10.1007/s11042-022-12447-5

This repository provides a pipeline for lane segmentation using a UNet model trained on the BDD100K dataset.

Standardized evaluation of modern object detectors beyond COCO.

A data visualization tool for the Berkley Deep Drive Dataset (available as a Plotly-Dash webapp or TKinter GUI app)

The official code open source version of BFDA - based on YOLOv5

Object detection model for BDD100K

Final project for the UPC Postgrau in Artificial Intelligence & Deep Learning, showcasing a SELF-DRIVING demo built around lane detection and object detection models. Integrates LaneNet (with ENet backbone and BDD100K transforms), a custom Mask R-CNN segmentation pipeline, and Faster R-CNN with ResNet50-FPN.

Some benchmarks and easy-to-understand explanations for the BDD100K dataset in MaskFormer.