Questions tagged [computer-vision]
The goal of computer vision is an automatic extraction of information out of images. It includes methods for acquiring, processing, analyzing, and understanding images and high-dimensional data from the real world in order to produce numerical or symbolic information, e.g., in the forms of decisions. The image data can take many forms, such as video sequences, views from multiple cameras, or multi-dimensional data from a medical scanner.
255 questions
0
votes
0
answers
16
views
Estimate 3D location of a tracked point with Known Camera parameters and Distance
I want to estimate the 3D point of X̅₂,
X̅₂ = [ X₂ Y₂ Z₂ ],
by using known variables:
X̅₁ is at world origin,
camera intrinsics K,
camera extrinsics w.r.t. X̅₁, namely [R₁ ∣ t₁],
...
- 1
0
votes
0
answers
23
views
ROS2 Node runs using python3 command but not with ros2 run
I am using a YOLOv5 package I found online (https://drive.google.com/drive/folders/1xC-gFmWQybQGagxtu8WxGqpFS-1Ek-V2) for my project. The thing is, the tutorial says to run the detection code by going ...
- 1
0
votes
0
answers
13
views
Making a ROS Node for a Semantic Segmentation Model I have devleoped
I have developed a SemSeg model using UNET. I want to implement it in my stack. I have converted the model to an ONNX version. I want to run it using a NVIDIA ORIN Device. What are the steps involved ...
- 1
0
votes
0
answers
17
views
How to calculate the twist of an object B with respect to frame C if I know the twist of another object B rigidly attached to A
the problem I have is the following. I have a fixed reference frame in the world (C) that does not move from its place and I have two objects (A and B) that are moving and are rigidly joined to each ...
1
vote
1
answer
49
views
SIM(3) conversion to SE(3) after pose graph optimization
To address scale drift in mono slam solutions, the euclidean transformations are converted to SIM(3) transformations and pose graph optimization is run on the 7dof pose graph.
My confusion is, how do ...
- 395
0
votes
0
answers
24
views
Get 3D boundingbox with 2D boundingbox and LiDAR data
I'm trying to create 3D bounding boxes from 2D bounding boxes created with yolov5 and point cloud data from a lidar sensor.I'm working with ROS2 and Gazebo and as a reference I'm trying to use this ...
0
votes
0
answers
17
views
Sensor Synchronization, IMU and Bundle Adjustment
Sensor synchronization is a general issue with SLAM systems, the ideal is having an IMU hardware synchronized with cameras.
There are a couple other possible variations though. One variation is ...
- 395
1
vote
0
answers
39
views
Covariance Matrix for the Measurement Model using a Camera
I'm currently taking an introductory course in Autonomous Systems where our project involves implementing the FastSLAM algorithm for an Alphabot2. This project uses a camera and Aruco markers as ...
- 11
0
votes
0
answers
9
views
Drawbacks of image rectification for 3d reconstruction
When running 3d scene reconstruction algorithms using monocular camera systems you can choose to rectify each pair of images or do 3d reconstruction with unrectified pairs using epipolar geometry.
...
- 395
0
votes
0
answers
7
views
Confusion on optimizing variable time-offset for VIO SLAM
I was reading the following paper on how you can incorporate variable time offset for a camera in VIO:
https://openaccess.thecvf.com/content_ECCV_2018/papers/Yonggen_Ling_Modeling_Varying_Camera-...
- 395
1
vote
0
answers
48
views
Ros2 drone camera projection issue
a few days ago I started working on a project whose aim is to avoid obstacles in the simulation autonomously by the drone and travel from point A to point B. Now I have reached the stage where I want ...
0
votes
0
answers
14
views
Calculate disparity for matching feature points
If I have matched a pair of feature points, could I calculate its disparity by simply subtract the u coordinates(uL - uR)? And could use this disparity to calculate the depth for this 3d point?
...
- 3
0
votes
0
answers
16
views
What post processing method was used to generate depth image as shown in the given example?
The 32FN method is a fast and effective method to estimate surface normal from a structured depth image. It appears the authors uses a "in-painting" method (please correct me if I am wrong) ...
0
votes
1
answer
61
views
Data sent over ROS2 topic is not reliable
Hello everyone of this wonderful forum. I am currently doing my first ROS2 project in which i have send data of aruco id that includes its [id number,x,y,z,orientation] data. I am sending this data ...
- 13
1
vote
0
answers
148
views
Does ORB-Slam3 perform better than ORB-Slam2?
I was trying to compare ORB-Slam2 performace with ORB-Slam3 performance. Consider following comparison from ORB-Slam3 paper (columns with numbers correspond to different sequences from EuRoC dataset):
...
- 151
0
votes
1
answer
154
views
Calibration of camera using ROS, without camera info
As part of a college project, I need to communicate with a TM5-700 robotic arm developed by Omron, and implement, for now, a pick and place pipeline. For that, I am required to access the camera ...
1
vote
3
answers
446
views
Installing Ros2 on the Jetson Xavier Nano
My team and I are trying to install Ros2 on the Jetson Xavier Nano so or robot can have computer vision, but the Jetson runs an old version of Ros. How can newer versions of ROS be used on the Jetson ...
- 11
0
votes
1
answer
50
views
Calculate target orientation
I have the following setup:
I want to move the "Drill" frame to "Target1" with the opposite Z axis direction. I have calculated translation between "Target1" and "...
0
votes
0
answers
62
views
Gazebo Garden : Green line in camera image with OGRE 1 render engine but not with OGRE 2
While loading the same map in gazebo garden with ogre 1 or ogre 2, the map looks the same in the GUI, but a green line appears in the camera image when using ogre1 (no distortion or noise applied, ...
- 1
1
vote
0
answers
86
views
World to pixel transformation in Pyrender
I'm trying to transform a point in a 3D world rendered with pyrender to pixel coordinates. The world to camera frame transformation seems to work, however the camera to pixel frame transformation is ...
- 11
1
vote
2
answers
130
views
Detecting door blockage using computer vision
I want to detect door blockage on a camera. Basically if the exit door is blocked by an object, it detects it as an anomaly. How can we do it? Is it possible to do it using OpenCV?
Remember, it doesn’...
- 111
1
vote
1
answer
101
views
Move gantry system to detected position in image plane
I have the following simplified gantry system (cartesian robot) that can move in the x axis direction:
My goal is to position the red pointer on top of the target (i.e., I want the real world x-...
0
votes
0
answers
47
views
Python How to fuse road segmentation data from an image to AOA data(distance, angle)
I have road segmentation data which is highlighted by the semi-transparent blue color.
It is an Numpy matrix with 0 means not road, 1 represents road.
I also have AOA data which tells the distance and ...
- 101
0
votes
0
answers
30
views
Making an automated chessboard
I want to create an automated chessboard. At the moment, I'm making the project and deciding what technology to use to detect which piece is at what position and how to move them. I'm not so skilled ...
- 1
3
votes
1
answer
82
views
How is the Homography formula derived?
In the Wikipedia article about Homography, the projective transformation between picture A to picture B is defined as:
$H_{ab} = R - \frac{t n^T}{d}$
However, the formula lacks the derivation. How is ...
0
votes
0
answers
71
views
Stereo Bundle Adjustment
When doing stereo based visual odometry and bundle adjustment how does the stereo constraint get factored into the photometric error?
Do you use the fact the second cameras location is known wrt the ...
- 395
0
votes
1
answer
83
views
Question about bundle adjustment
When doing bundle adjustment do landmarks gets optimized as stand alone variables or are they viewed as a function of the robots first position when they are detected?
- 395
0
votes
1
answer
291
views
Why would you use direct SLAM over feature-based?
I am curious, why would you ever choose direct slam(photometric consistency, etc) over feature based slam? Particularly for dynamic scenes?
It seems you're always better off aligning fewer points ...
- 395
0
votes
1
answer
40
views
Stabilising 3d cameras for mapping outdoors - mechanical or programmatic
I am preparing to mount Intel Realsense and Zed and Oak-d cameras onto a 4 wheel mobility scooter to do mapping of my local area.
I'm not looking for recommendations but rather an understanding of ...
- 130
0
votes
0
answers
303
views
IMU to Camera transformation
Suppose you are implementing SLAM or VIO system, you have your camera C and IMU and a coordinate transformation between them (R, T).
How do convert IMU measurements to camera accelerations and ...
- 395
1
vote
1
answer
196
views
Calculating the depth in a general stereo vision system
When calculating the depth of a pixel using two cameras, most of the literature I have read describes stereo vision with the cameras having parallel optical axis as shown in the picture below. Because ...
- 113
2
votes
1
answer
316
views
Pipeline for dense reconstruction using pose estimation from orb-slam and stereo camera
I am trying to perform dense reconstruction using a sequence of images from a calibrated stereo camera. I have been using orb-slam3 to give me the camera's pose estimation. I am also generating the ...
- 23
0
votes
1
answer
344
views
How to perform time-offset calibration use GTSAM in Camera-IMU system?
Assume we have camera poses data and IMU data(which include gyr and acc data) with different timestamps, now I want to estimate this value, I known that this can be done in many ways(EKF,MSCKF), but I ...
- 1
0
votes
1
answer
91
views
Camera Calibration Matrix accuracy
When we do camera calibration, we have to find calibration matrix $M$, which is found by first finding extrinsic matrix and then intrinsic matrix. To validate the accuracy of calibration matrix, we ...
- 105
0
votes
1
answer
106
views
6D pose estimation of a known 3D CAD object [closed]
I'm looking for a codebase for 6DOF pose estimation of a known 3D CAD object with RGB or RGBD. It must be:
-Usable commercially (licensed under BSD, MIT, BOOST, etc.), not GPL.
-Easy to setup and use (...
- 248
0
votes
1
answer
113
views
2
votes
0
answers
70
views
Monocular SLAM position estimation
I have been reading about mono SLAM and I have a couple questions which are not so clear:
When doing pose estimation, I have seen people both reference homography transformation and also reference 5\...
- 395
0
votes
0
answers
43
views
How to find fundamental matrix based on other fundamental matrix and camera movement?
I am trying to speed up some multi-camera system that relies on calculation of fundamental matrices between each camera pair.
Please notice the following is pseudocode. ...
- 101
0
votes
1
answer
99
views
Center line steering with differential drive
I’ve got 2 motors that drives my unit and I can easily steer it manually by simply adding more speed to one wheel vs the other.
Now I’m automating the rover using image processing to follow a driveway....
- 23
0
votes
1
answer
76
views
How does baseline work with forward motion in Monocular Visual Odometry
In my computer vision class we are studying visual odometry. In this we calculate the Fundamental matrix using 2 frames from forward motion using the concept of stereo-vision. But in stereo vision the ...
- 1
1
vote
2
answers
795
views
roslaunch exit code 2 error
I am implementing a simple roslaunch file but I am getting an exit code 2 status I have pasted the exact log below any idea what would cause this and how I can rectify it.
...
- 21
0
votes
1
answer
227
views
Mobile robotic arm camera position
I'm working on my graduation project which is a mobile robotic arm that performs pick and place tasks.
I wanted to know the possible ways to mount the camera where i can reach the desired destination ...
- 1
0
votes
1
answer
33
views
Explanation of the epipolar constraint in stereo imaging
I am watching this video, in which the epipolar constraint is defined as
$x_l \cdot (x_l \times t)=0$
It means that the vector $x_l$ that passes through the observed point and the left camera origin ...
- 101
0
votes
1
answer
24
views
What are the usual results for pedestrian posture detection?
When talking about lane detection in a road, usually robotic computer vision system give lines in an image.
For detecting obstacles, those system give Bounding Boxes.
My question is, for pedestrian (...
- 103
3
votes
1
answer
327
views
Unstable principal point estimation in zoom camera calibration
I have a camera with a 1/2.9" image sensor from FLIR and a 12-50mm motorized zoom lens from Theia. Because the lens is a motorized zoom lens, I tried to calibrate the camera for different zoom ...
- 31
0
votes
2
answers
1k
views
Number of pixels required to identify something
When using an RGB camera, what is the minimum number of pixels of an object that you need to identify something (theoretical or rule of thumb)?
The image would be taken out in the wild with a complex ...
1
vote
1
answer
44
views
Can you plug a CMOS camera (like the iPhone 8 back camera) into an mPCIe in a laptop?
I read that 12 megapixel iPhone cameras don’t record video in 12 megapixels because the processors are not able to keep up. It seems though, that if I could plug the CMOS sensor to the CPU of a laptop,...
- 113
1
vote
0
answers
166
views
Cheap global shutter cameras with external synchronization
Please recommend cameras for multi-camera setup satisfying the following conditions:
Relatively cheap (up to 100$ per camera).
Global shutter.
No filters on lenses or pixels.
Resolution of 1 MP or ...
- 111
1
vote
1
answer
51
views
Autonomous harvesting robots. What are model possible components and interactions for the entire system? [closed]
I found a video which talk about best autonomous robots that use IA technology to harvest crops. The video is very interesting because, the “ecosystem” of each robot varies a lot from company to ...
- 11
2
votes
1
answer
220
views
Depth camera pixel values to 3D coordinates
I have a robotic arm 6-DOF and an Intel RealSense depth camera. I want to go from pixel values to 3D world coordinates, is it possible with eye-in-hand calibration? If yes can you guide me on how to ...
