Name		Name	Last commit message	Last commit date
parent directory ..
IO_places365.txt		IO_places365.txt
LICENSE		LICENSE
README.md		README.md
W_sceneattribute_wideresnet18.npy		W_sceneattribute_wideresnet18.npy
cam.jpg		cam.jpg
categories_places365.txt		categories_places365.txt
input.jpg		input.jpg
input2.jpg		input2.jpg
labels_sunattribute.txt		labels_sunattribute.txt
places365.py		places365.py

README.md

Places365

Input

from https://github.com/CSAILVision/places365

Ailia input shape: (1, 3, 224, 224)

Output

If model type is resnet18, resnet50 or alexnet, result is as below.

--SCENE CATEGORIES:
	0.616 -> patio
	0.314 -> restaurant_patio
	0.020 -> beer_garden
	0.011 -> courtyard
	0.011 -> porch

If model type is wideresnet18, result is as below and class activation map is generated.

--TYPE OF ENVIRONMENT:
	outdoor
--SCENE CATEGORIES:
	0.651 -> patio
	0.068 -> restaurant_patio
	0.043 -> porch
	0.026 -> courtyard
	0.022 -> picnic_area
--SCENE ATTRIBUTES:
	 man-made, no horizon, natural light, open area, foliage, wood, vegetation, leaves, trees

Class activation map

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 places365.py

If you want to specify the input image, put the image path after the --input option.

$ python3 places365.py --input IMAGE_PATH

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 places365.py --video VIDEO_PATH

You can select a model from resnet18 | resnet50 | alexnet | wideresnet18 by adding --model option.

In the case of wideresnet18 model, you can use --savepath option to save the class activation map file.

(ex1)$ python3 places365.py --model wideresnet18 --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

(ex2)$ python3 places365.py --model wideresnet18 --video VIDEO_PATH --savepath SAVE_VIDEO_PATH