module 'scenedetect.detectors' has no attribute 'get_available'

[ishandutta2007]

import scenedetect
scene_detectors = scenedetect.detectors.get_available()
smgr_content = scenedetect.manager.SceneManager(PySceneDetectArgs(input=video_path, type='content'), scene_detectors)

module 'scenedetect.detectors' has no attribute 'get_available'

scenedetect version is 0.5 and python version is 3.5

[Breakthrough]

Hi @ishandutta2007

I removed the get_available() method in v0.5 as it's not really needed to get all detection methods at runtime. This could also be achieved in a better way I think, if required, going forwards, however I'm still open to the older style.

Thoughts?

[ishandutta2007]

Can you quickly help me understand how would the above code now look like in v0.5?

[wjs018]

@ishandutta2007
You just add the detector directly. Here is a simplified example from one of my functions where I add a content detector. To see the whole thing, look at the analyze_video function here.

import scenedetect

video_mgr = scenedetect.VideoManager([video_file])    #put your filename in this line
stats_mgr = scenedetect.stats_manager.StatsManager()
scene_mgr = scenedetect.SceneManager(stats_mgr)

# Now add the content detector
scene_mgr.add_detector(scenedetect.ContentDetector(threshold=30, min_scene_len=10))

I haven't tried this, but you should be able to add multiple detectors with multiple calls to add_detector. Then, if any of your detectors triggers a scene break on a frame, it is included in the scene list. This would let you have both a ContentDetector and a ThresholdDetector (or your own custom detector) analyzing your video at the same time.

Edit: Looks like this function is where the SceneManager loops over the available detectors. So, multiple detectors shouldn't be a problem.

[ishandutta2007]

Well if you are seeking advices, I would say the newer approach is fine.

In fact as end user I would prefer the APIs to be as high level as possible, as much abstracted out as possible. I may not even care to expose any of StatsManager, VideoManager or SceneManager classes. Something like

import scenedetect
video = scenedetect.load(video_file_name)
scene_list = video.get_scenes(threshold=30, min_scene_len=10)
scene_list[2].write(scene_file_name)

Anyways whatever you are doing doesn't seem to be too out of the standards,
I am saying just from the point of view of a programmers ease of access.
Whatever you decide, just keep docs and examples updated, Cheers.