jetson_tello.app¶

jetson_tello.app.run_jetson_tello_app(fly, process_frame, drone=None, on_frame_decoded=None, wait_for_wifi=True)¶

Fly and control a drone while analysing video frames using CUDA.

This function takes two main arguments - an async function to control the drone in whatever way you want, and a callback called with video frame data ready and loaded into CUDA memory for analysis.

The process_frame callback always waits for the next available frame, skipping any that arrive while processing the last. Frame analysis can take as long as it needs without falling behind the live video.

There is also an optional on_frame_decoded callback which is called for every frame. This must be fast enough to not clog things up and cause latency.

Parameters

fly (Coroutine function which takes a tello_asyncio.tello.Tello drone argument.) – Awaitable function which controls the drone. The app exits when complete.
process_frame (Awaitable or plain function taking tello_asyncio.tello.Tello drone, jetson_tello.types.DecodedFrame frame, cudaImage cuda) – Callback called with frame data loaded into CUDA memory
drone (tello_asyncio.tello.Tello) – Optional drone instance. One will be created automatically if not provided
on_frame_decoded (Awaitable or plain function taking jetson_tello.types.DecodedFrame frame) – Optional callback called for every decoded frame
wait_for_wifi (bool) – If true, wait for connection to the drone’s WiFi network before proceeding (Linux only)

jetson_tello.coco¶

exception jetson_tello.coco.InvalidCocoClassError(value)¶

Bases: Exception

Exception raised if a COCO class could not identified from the given value.

jetson_tello.coco.get_coco_class(value)¶

COCO class from a class ID, name or object detection object.

Parameters: value – The class ID, name or object detection
Return type: jetson_tello.coco.classes.CocoClass
Throws: jetson_tello.coco.functions.InvalidCocoClassError

jetson_tello.coco.get_coco_class_by_id(id)¶

COCO class from a class ID.

Parameters: id (int) – The class ID
Return type: jetson_tello.coco.classes.CocoClass
Throws: jetson_tello.coco.functions.InvalidCocoClassError

jetson_tello.coco.get_coco_class_by_name(name)¶

COCO class from a class name.

Parameters: name (string) – The class name
Return type: jetson_tello.coco.classes.CocoClass
Throws: jetson_tello.coco.functions.InvalidCocoClassError

jetson_tello.video¶

class jetson_tello.video.DecodedFrame(number, width, height, data)

Bases: tuple

Video frame data and information

data: Decoded frame data as bytes

height: Frame image height in pixels

number: Position in the sequence of successfully decoded frames

width: Frame image width in pixels

class jetson_tello.video.H264DecoderAsync

Bases: object

Decodes a stream of h.264 encoded video frames, making them available for analysis.

async decode_frames(video_stream, on_frame_decoded=None)

Begin decoding video frames.

Parameters

video_stream (Asynchronous iterator of h.264 frames) – The video stream
on_frame_decoded (Awaitable or plain function taking jetson_tello.types.DecodedFrame) – Optional callback called for each successfully decoded frame. Must be fast!

property decoded_frame: The most recently decoded frame. :rtype: jetson_tello.types.DecodedFrame

exception jetson_tello.video.NoFrameData

Bases: Exception

Exception raised when no frame data is available, either because the frame is corrupted, or intentionally for reasons known only to the h.264 encoder.

jetson_tello.video.decoded_frame_to_cuda(decoded_frame)

Loads frame data into CUDA memory.

Parameters: decoded_frame (jetson_tello.types.DecodedFrame) –
Return type: cudaImage

jetson_tello.video.decoded_frame_to_numpy_array(decoded_frame)

Takes a decoded frame and returns a NumPy array of the RGB values.

Parameters: decoded_frame (jetson_tello.types.DecodedFrame) –
Return type: numpy.array

jetson_tello.video.h264_frame_to_cuda(h264_frame)

Decodes raw h.264 frame data and copies it into CUDA memory.

Parameters: frame (bytes) – The raw frame data
Return type: (jetson_tello.types.DecodedFrame, cudaImage)
Throws: jetson_tello.video.NoFrameData

jetson_tello.video.h264_frame_to_numpy_array(h264_frame)

Decodes raw h.264 frame data and copies it into a NumPy array ready for analysis.

Parameters: h264_frame – The h.264 encoded frame data
Return type: (jetson_tello.types.DecodedFrame, numpy.ndarray)
Throws: jetson_tello.video.NoFrameData