object detection / instance segmentation
Matterport's implementation of Mask R-CNN for object detection and instance segmentation using Keras and TensorFlow.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Transformer-based unified detection/segmentation framework; choose for state-of-the-art accuracy and joint tasks, but requires PyTorch.
[CVPR 2023] Official implementation of the paper "Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation"
Full-featured PaddlePaddle toolkit for detection/segmentation; choose for active development, deployment, and PaddlePaddle compatibility.
Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.
PyTorch toolbox with huge model zoo; choose for model variety and community support, but requires PyTorch.
OpenMMLab Detection Toolbox and Benchmark
Anchor-free real-time instance segmentation; choose for faster inference than Mask R-CNN, but requires PyTorch.
[CVPR 2020] CenterMask : Real-Time Anchor-Free Instance Segmentation
Anchor-free fully convolutional instance segmentation; choose for simplicity and no proposal stage, but requires PyTorch.
SOLO and SOLOv2 for instance segmentation, ECCV 2020 & NeurIPS 2020.
Transformer-based universal image segmentation; choose for state-of-the-art panoptic/instance segmentation, but heavy runtime.
Code release for "Masked-attention Mask Transformer for Universal Image Segmentation"
Up-to-date anchor-free instance segmentation; choose for real-time performance and active fixes.
[CVPR 2020] CenterMask : Real-time Anchor-Free Instance Segmentation
Modern transformer-based detection/segmentation model; choose for real-time speed and COCO accuracy, but it is a specific architecture.
RF-DETR is a real-time object detection and segmentation model architecture developed by Roboflow, SOTA on COCO, designed for fine-tuning. [ICLR 2026]
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Previous-generation Mask Scoring R-CNN research code; choose for its mask-scoring head, but it is unmaintained.
Codes for paper "Mask Scoring R-CNN".
Previous-generation object detector without segmentation; choose only for legacy detection research.
Fast R-CNN
Older instance-segmentation network enhancing Mask R-CNN with path aggregation; choose for PANet architecture, but no active support.
PANet for Instance Segmentation and Object Detection
Previous-generation lightweight detection/segmentation framework; choose if you need a simpler API, but it is stale and limited.
A Simple and Versatile Framework for Object Detection and Instance Recognition
Medical-domain framework including Mask R-CNN; choose for 2D/3D medical imaging tools, but it is domain-specific.
The Medical Detection Toolkit contains 2D + 3D implementations of prevalent object detectors such as Mask R-CNN, Retina Net, Retina U-Net, as well as a training and inference framework focused on dealing with medical images.
Original Caffe2 research platform for Mask R-CNN; choose for legacy Caffe2 deployment, but it is archived.
FAIR's research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet.
TensorFlow implementation of Faster R-CNN, detection only; choose for legacy TF detection without segmentation.
Tensorflow Faster RCNN for Object Detection
Original Matlab Faster R-CNN; choose only for reproducing the paper, not for practical use.
Faster R-CNN
Center-based object detector with pose estimation; choose for detection/pose tasks, but no instance segmentation.
Object detection, 3D detection, and pose estimation using center point detection:
Does the same job, but ships as an app. Useful to a person, not swappable into a codebase.
High-level wrapper for segmentation models; choose for quick prototyping with simple API, but no training flexibility.
Visit PixelLib's official documentation https://pixellib.readthedocs.io/en/latest/
These projects were analysed and named Mask_RCNN among their alternatives. The relationship is not symmetric — how Mask_RCNN rates them is a separate judgement, made when Mask_RCNN is analysed in its own right.
calls Mask_RCNN “Mask R-CNN in Keras/TensorFlow, choose for TF workflows.”
OpenMMLab Detection Toolbox and Benchmark
calls Mask_RCNN “Keras/TensorFlow Mask R-CNN — choose for TensorFlow deployment of that single model.”
Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.