Yet-Another-EfficientDet-Pytorch
simple-faster-rcnn-pytorch
Our great sponsors
Yet-Another-EfficientDet-Pytorch | simple-faster-rcnn-pytorch | |
---|---|---|
1 | 1 | |
5,183 | 3,891 | |
- | - | |
0.0 | 0.0 | |
over 2 years ago | almost 3 years ago | |
Jupyter Notebook | Jupyter Notebook | |
GNU Lesser General Public License v3.0 only | GNU General Public License v3.0 or later |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
Yet-Another-EfficientDet-Pytorch
-
Bounding box annotations and object orientation
However, there are papers on oriented object detectors (see https://arxiv.org/pdf/1911.07732.pdf) for example. In that paper, they do achieve better results using oriented bounding boxes. If you want to go down that route, I would suggest using the EfficientDet model, because the PyTorch code that you'll find for it is quite easy to understand and modify. For example, I've taken https://github.com/zylo117/Yet-Another-EfficientDet-Pytorch, and modified it to include a "thing-ness" logit, and this was pretty easy to do. Classic EfficientDet models only include logits (aka output neurons that get softmax-ed) for each class, and if any one of these class neurons is greater than 0.5, then it is considered "a thing". Anyway - that's digression, but my point is that I've thought about adding oriented box support to an EfficientDet model, and it didn't seem to be too hard, although I haven't actually done it. If I was to start now, I would probably go with https://github.com/rwightman/efficientdet-pytorch, since Ross Wightman's models are becoming a de-facto standard in the PyTorch world for all things image-related.
simple-faster-rcnn-pytorch
-
ISO Easy to Modify and Use Faster RCNN PyTorch Implementation
Hi all, as the title suggests I'm looking for a GitHub repo where I can edit a Faster RCNN implementation rather easily. I'm basically looking to test an idea where I have multiple branches with feature map and bounding boxes as inputs. I've modified the built-in torchvision implementation once before, but I think it's a little more complicated than I like, and I'd rather not release the entire torchvision package as part of my own work in the future. I have looked briefly into this repo https://github.com/chenyuntc/simple-faster-rcnn-pytorch/blob/master/trainer.py but it only supports a batch size of 1, and I'm not sure what it'd take to expand that capability. Is there anything better out there?
What are some alternatives?
darknet - YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
mmdetection - OpenMMLab Detection Toolbox and Benchmark
efficientdet-pytorch - A PyTorch impl of EfficientDet faithful to the original Google impl w/ ported weights
Real-time-Object-Detection-for-Autonomous-Driving-using-Deep-Learning - 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.
fashionpedia-api - Python API for Fashionpedia Dataset
automl - Google Brain AutoML
TFLiteDetection - TensorFlow Lite Object Detection Python Implementation
OwnPhotos - Self hosted alternative to Google Photos
SipMask - SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation (ECCV2020)
yolov3-tf2 - YoloV3 Implemented in Tensorflow 2.0
SynthDet - SynthDet - An end-to-end object detection pipeline using synthetic data
opencv - Experimenting using Machine Vision OpenCV and Python to create software suitable for driving a Golf launch monitor similar to technology like SkyTrak, GC2 and GC Quad