/maskrcnn-benchmark

Fast, modular reference implementation of Instance Segmentation and Object Detection algorithms in PyTorch.

Primary LanguagePythonMIT LicenseMIT

Faster R-CNN and Mask R-CNN in PyTorch 1.0

This project aims at providing the necessary building blocks for easily creating detection and segmentation models using PyTorch 1.0.

alt text

Highlights

  • PyTorch 1.0: RPN, Faster R-CNN and Mask R-CNN implementations that matches or exceeds Detectron accuracies
  • Very fast: up to 2x faster than Detectron and 30% faster than mmdetection during training. See MODEL_ZOO.md for more details.
  • Memory efficient: uses roughly 500MB less GPU memory than mmdetection during training
  • Multi-GPU training and inference
  • Batched inference: can perform inference using multiple images per batch per GPU
  • CPU support for inference: runs on CPU in inference time. See our webcam demo for an example
  • Provides pre-trained models for almost all reference Mask R-CNN and Faster R-CNN configurations with 1x schedule.

Webcam and Jupyter notebook demo

We provide a simple webcam demo that illustrates how you can use maskrcnn_benchmark for inference:

cd demo
# by default, it runs on the GPU
# for best results, use min-image-size 800
python webcam.py --min-image-size 800
# can also run it on the CPU
python webcam.py --min-image-size 300 MODEL.DEVICE cpu
# or change the model that you want to use
python webcam.py --config-file ../configs/caffe2/e2e_mask_rcnn_R_101_FPN_1x_caffe2.yaml --min-image-size 300 MODEL.DEVICE cpu
# in order to see the probability heatmaps, pass --show-mask-heatmaps
python webcam.py --min-image-size 300 --show-mask-heatmaps MODEL.DEVICE cpu

A notebook with the demo can be found in demo/Mask_R-CNN_demo.ipynb.

Installation

Check INSTALL.md for installation instructions.

Model Zoo and Baselines

Pre-trained models, baselines and comparison with Detectron and mmdetection can be found in MODEL_ZOO.md

Inference in a few lines

We provide a helper class to simplify writing inference pipelines using pre-trained models. Here is how we would do it. Run this from the demo folder:

from maskrcnn_benchmark.config import cfg
from predictor import COCODemo

config_file = "../configs/caffe2/e2e_mask_rcnn_R_50_FPN_1x_caffe2.yaml"

# update the config options with the config file
cfg.merge_from_file(config_file)
# manual override some options
cfg.merge_from_list(["MODEL.DEVICE", "cpu"])

coco_demo = COCODemo(
    cfg,
    min_image_size=800,
    confidence_threshold=0.7,
)
# load image and then run prediction
image = ...
predictions = coco_demo.run_on_opencv_image(image)

Perform training on COCO dataset

For the following examples to work, you need to first install maskrcnn_benchmark.

You will also need to download the COCO dataset. We recommend to symlink the path to the coco dataset to datasets/ as follows

We use minival and valminusminival sets from Detectron

# symlink the coco dataset
cd ~/github/maskrcnn-benchmark
mkdir -p datasets/coco
ln -s /path_to_coco_dataset/annotations datasets/coco/annotations
ln -s /path_to_coco_dataset/train2014 datasets/coco/train2014
ln -s /path_to_coco_dataset/test2014 datasets/coco/test2014
ln -s /path_to_coco_dataset/val2014 datasets/coco/val2014

You can also configure your own paths to the datasets. For that, all you need to do is to modify maskrcnn_benchmark/config/paths_catalog.py to point to the location where your dataset is stored. You can also create a new paths_catalog.py file which implements the same two classes, and pass it as a config argument PATHS_CATALOG during training.

Single GPU training

Most of the configuration files that we provide assume that we are running on 8 GPUs. In order to be able to run it on fewer GPUs, there are a few possibilities:

1. Run the following without modifications

python /path_to_maskrnn_benchmark/tools/train_net.py --config-file "/path/to/config/file.yaml"

This should work out of the box and is very similar to what we should do for multi-GPU training. But the drawback is that it will use much more GPU memory. The reason is that we set in the configuration files a global batch size that is divided over the number of GPUs. So if we only have a single GPU, this means that the batch size for that GPU will be 8x larger, which might lead to out-of-memory errors.

If you have a lot of memory available, this is the easiest solution.

2. Modify the cfg parameters

If you experience out-of-memory errors, you can reduce the global batch size. But this means that you'll also need to change the learning rate, the number of iterations and the learning rate schedule.

Here is an example for Mask R-CNN R-50 FPN with the 1x schedule:

python tools/train_net.py --config-file "configs/e2e_mask_rcnn_R_50_FPN_1x.yaml" SOLVER.IMS_PER_BATCH 2 SOLVER.BASE_LR 0.0025 SOLVER.MAX_ITER 720000 SOLVER.STEPS "(480000, 640000)" TEST.IMS_PER_BATCH 1

This follows the scheduling rules from Detectron. Note that we have multiplied the number of iterations by 8x (as well as the learning rate schedules), and we have divided the learning rate by 8x.

We also changed the batch size during testing, but that is generally not necessary because testing requires much less memory than training.

Multi-GPU training

We use internally torch.distributed.launch in order to launch multi-gpu training. This utility function from PyTorch spawns as many Python processes as the number of GPUs we want to use, and each Python process will only use a single GPU.

export NGPUS=8
python -m torch.distributed.launch --nproc_per_node=$NGPUS /path_to_maskrcnn_benchmark/tools/train_net.py --config-file "path/to/config/file.yaml"

Abstractions

For more information on some of the main abstractions in our implementation, see ABSTRACTIONS.md.

Adding your own dataset

This implementation adds support for COCO-style datasets. But adding support for training on a new dataset can be done as follows:

from maskrcnn_benchmark.structures.bounding_box import BoxList

class MyDataset(object):
    def __init__(self, ...):
        # as you would do normally
    
    def __getitem__(self, idx):
        # load the image as a PIL Image
        image = ...

        # load the bounding boxes as a list of list of boxes
        # in this case, for illustrative purposes, we use
        # x1, y1, x2, y2 order.
        boxes = [[0, 0, 10, 10], [10, 20, 50, 50]]
        # and labels
        labels = torch.tensor([10, 20])

        # create a BoxList from the boxes
        boxlist = BoxList(boxes, image.size, mode="xyxy")
        # add the labels to the boxlist
        boxlist.add_field("labels", labels)

        if self.transforms:
            image, boxlist = self.transforms(image, boxlist)

        # return the image, the boxlist and the idx in your dataset
        return image, boxlist, idx

    def get_img_info(self, idx):
        # get img_height and img_width. This is used if
        # we want to split the batches according to the aspect ratio
        # of the image, as it can be more efficient than loading the
        # image from disk
        return {"height": img_height, "width": img_width}

That's it. You can also add extra fields to the boxlist, such as segmentation masks (using structures.segmentation_mask.SegmentationMask), or even your own instance type.

For a full example of how the COCODataset is implemented, check maskrcnn_benchmark/data/datasets/coco.py.

Note:

While the aforementioned example should work for training, we leverage the cocoApi for computing the accuracies during testing. Thus, test datasets should currently follow the cocoApi for now.

Troubleshooting

If you have issues running or compiling this code, we have compiled a list of common issues in TROUBLESHOOTING.md. If your issue is not present there, please feel free to open a new issue.

License

maskrcnn-benchmark is released under the MIT license. See LICENSE for additional details.