D435i camera high-speed mode save depth data to .npz files

[Ligcox]

errorexample.txt

---

Required Info
Camera Model	D435i
Firmware Version	05.12.13.50
Operating System & Version	Win (8.1/10)
Kernel Version (Linux Only)	-
Platform	PC
SDK Version	2.25.0
Language	python
Segment	others

Issue Description

I tried to use the D435i high-speed mode(https://dev.intelrealsense.com/docs/high-speed-capture-mode-of-intel-realsense-depth-camera-d435) to record the image data and write them to the .npz file. The program ran into problems when I tried to add data to a variable in the for loop. I tried the following

1. Comment an invalid operation statement, such as using OpenCV to display colormap
2. Load data with an empty list and add data to the list using methods append or insert, for example, res=[]; res.append(depth_image)
3. Read the image data through two separate threads and add it to the final result
4. Create a FIFO queue using queue.Queue() and return the res
5. Change the different wait times in pipeline.wait_for_frames()

Run into the same problem using these methods, they're similar

Traceback (most recent call last):
File ".\src\realsense\save_data.py", line 17, in
frames = pipeline.wait_for_frames()
RuntimeError: Frame didn't arrive within 5000

I made sure that my RealSense was connected to the computer correctly, and I used the original data cable. RealSense work fine in the Intel RealSense Viewer and are displayed over the USB3.2 cable connection.

When I comment out the code that saves the image, the error goes away (for example, comment line 28 of the instance file). This error occurs during the 15th for loop, and you can see through the debuger that the program works normally during the previous loops.

The error demonstration code is as follows

    import pyrealsense2 as rs
    import cv2
    import numpy as np
    import time
    import tqdm
    
    pipeline = rs.pipeline()
    config = rs.config()
    config.enable_stream(rs.stream.depth, 0, 848, 100, rs.format.z16, 300)
    pipeline.start(config)
    
    cv2.namedWindow('RealSense', cv2.WINDOW_AUTOSIZE)
    num = 1000
    res = [None for i in range(num)]
    
    for i in tqdm.tqdm(range(num)):
        frames = pipeline.wait_for_frames()
        depth_frame = frames.get_depth_frame()
        if not depth_frame:
            continue
    
        depth_image = np.asanyarray(depth_frame.get_data())
        depth_colormap = cv2.applyColorMap(cv2.convertScaleAbs(
            depth_image, alpha=0.085), cv2.COLORMAP_JET)
        cv2.imshow('RealSense', depth_colormap)
        cv2.waitKey(1)
    
        res[i] = depth_image
    
    cv2.destroyAllWindows()
    np.savez_compressed("realsense", *res)

[MartyG-RealSense]

Hi @Ligcox When an append operation in a Python script in librealsense fails after the 15th frame, it is typically due to a limitation on saving frames into a collection.

#946 (comment)

A solution that has worked for RealSense users who experience this issue is to store the frames in memory using the SDK's Keep() function. The links below have examples of Python code for Keep():

#3164 (comment)

#3121 (comment)

[MartyG-RealSense]

Hi @Ligcox Do you require futher assistance with this case, please? Thanks!

[MartyG-RealSense]

Case closed due to no further comments received.