Skip to content

minjung-mmc/SFND_Camera_Final

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
Reference
Reference
 
 
dat
dat
 
 
images
images
 
 
src
src
 
 
CMakeLists.txt
CMakeLists.txt
 
 
README.md
README.md
 
 

Repository files navigation

SFND_3D_Object_Tracking

FP.1 Match 3D Objects

match the current and previous boundingbox that share the largest number of keypoints, and return index of boundingbox to std::map<int, int> &bbBestmatches.

void matchBoundingBoxes(std::vector<cv::DMatch> &matches, std::map<int, int> &bbBestMatches, DataFrame &prevFrame, DataFrame &currFrame)
{
    // match(queryIdx = prev, trainIdx = curr)
    cv::Mat matrix(currFrame.boundingBoxes.size(),prevFrame.boundingBoxes.size(),cv::DataType<int>::type, cv::Scalar(0));

    for (auto it0 = matches.begin(); it0 != matches.end(); ++it0)
    {
        int currIdx = -1, prevIdx = -1;

        cv::Point currPt = currFrame.keypoints[it0->trainIdx].pt;
        for (auto it1 = currFrame.boundingBoxes.begin(); it1 != currFrame.boundingBoxes.end(); ++it1)
        {            
            if (it1->roi.contains(currPt) == true)
            {
                currIdx = it1->boxID;
                break;
            }
        }
        if(currIdx < 0) continue;

        cv::Point prevPt = prevFrame.keypoints[it0->queryIdx].pt;
        for (auto it2 = prevFrame.boundingBoxes.begin(); it2 != prevFrame.boundingBoxes.end(); ++it2)
        {
            if (it2->roi.contains(prevPt) == true)
            {
                prevIdx = it2->boxID;
                break;
            }
        }

        if (prevIdx >= 0)
        {
            ++(matrix.at<int>(currIdx, prevIdx));
        }
    }
    
    // Find best match (prevBoxIdx, currentBoxIdx)
    if (matrix.rows < matrix.cols)
    {
        for (int r = 0; r < matrix.rows; ++r)
        {
            int maxIdx = 0;
            for (int c = 0; c < matrix.cols; ++c)
            {
                if (matrix.at<int>(r, c) > matrix.at<int>(r, maxIdx))
                {
                    maxIdx = c;
                }
            }
            bbBestMatches.insert(pair<int, int>(maxIdx, r));
        }
    }
    else
    {
        for (int c = 0; c < matrix.rows; ++c)
        {
            int maxIdx = 0;
            for (int r = 0; r < matrix.cols; ++r)
            {
                if (matrix.at<int>(r, c) > matrix.at<int>(maxIdx, c))
                {
                    maxIdx = r;
                }
            }
            bbBestMatches.insert(pair<int, int>(c, maxIdx));
        }
    }
}

FP.2 Compute Lidar-based TTC

Compute the time-to-collision for all matched 3D objects based on Lidar measurements alone. Implement the estimation in a way that makes it robust against outliers which might be way too close and thus lead to faulty estimates of the TTC. -> I found Lidar points within ego line, and computed median value of those lidar points to estimate TTC.

void computeTTCLidar(std::vector<LidarPoint> &lidarPointsPrev,
                     std::vector<LidarPoint> &lidarPointsCurr, double frameRate, double &TTC)
{
    double laneWidth = 4.0;
    double dT = 1 / frameRate;

    // find Lidar points within ego lane
    vector<double> Xprev, Xcurr;
    for (auto it = lidarPointsPrev.begin(); it != lidarPointsPrev.end(); ++it)
    {
        if (abs( it->y ) <= laneWidth/2.0 )
        {
            Xprev.push_back(it->x);
        }
    }
    for (auto it = lidarPointsCurr.begin(); it != lidarPointsCurr.end(); ++it)
    {
        if (abs( it->y ) <= laneWidth/2.0 )
        {
            Xcurr.push_back(it->x);
        }
    }

    // compute median value of Lidar point X
    std::sort(Xprev.begin(), Xprev.end());
    int medIndexPrev = floor(Xprev.size() / 2.0); 
    double medXprev = Xprev.size() % 2 == 0 ? (Xprev[medIndexPrev - 1] + Xprev[medIndexPrev]) / 2.0 : Xprev[medIndexPrev];

    std::sort(Xcurr.begin(), Xcurr.end());
    int medIndexCurr = floor(Xcurr.size() / 2.0); 
    double medXcurr = Xcurr.size() % 2 == 0 ? (Xcurr[medIndexCurr - 1] + Xcurr[medIndexCurr]) / 2.0 : Xcurr[medIndexCurr]; 
    
    // compute TTC
    TTC = medXcurr * dT / (medXprev - medXcurr);
}

FP.3 Associate Keypoint Correspondences with Bounding Boxes

Find all keypoint matches that belong to each 3D object. You can do this by simply checking whether the corresponding keypoints are within the region of interest in the camera image. Compute a robust mean of all the euclidean distances between keypoint matches and then remove those that are too far away from the mean. Store the result in the "kptMatches" property of the respective bounding box. -> I used bottom 80% of the euclidean distance, and assumed the rest as outliers.

void clusterKptMatchesWithROI(BoundingBox &boundingBox, std::vector<cv::KeyPoint> &kptsPrev, std::vector<cv::KeyPoint> &kptsCurr, std::vector<cv::DMatch> &kptMatches)
{
    std::vector<double> euclidDistance;
    std::vector<cv::DMatch> boxedMatches;
    double outlierRatio = 0.2;

    // Find matches inside the bounding box
    for (auto it = kptMatches.begin(); it != kptMatches.end(); ++it)
    {
        if (boundingBox.roi.contains(kptsCurr[it->trainIdx].pt) == true)
        {
            euclidDistance.push_back(cv::norm(cv::Mat(kptsCurr[it->trainIdx].pt), cv::Mat(kptsPrev[it->queryIdx].pt)) );
        }
    }

    sort(euclidDistance.begin(), euclidDistance.end());

    for (auto it = kptMatches.begin(); it != kptMatches.end(); ++it)
    {
        double distance;
        if (boundingBox.roi.contains(kptsCurr[it->trainIdx].pt) == true)
        {
            distance = cv::norm(cv::Mat(kptsCurr[it->trainIdx].pt), cv::Mat(kptsPrev[it->queryIdx].pt));

            if (distance <= threshold)
            {
                boundingBox.kptMatches.push_back(*it);
                boundingBox.keypoints.push_back(kptsCurr[it->trainIdx]);
            }
        }
    }
}

FP.4 Compute Camera-based TTC

Compute the time-to-collision for all matched 3D objects based on Camera measurements. Implement the estimation in a way that makes it robust against outliers which might be way too close and thus lead to faulty estimates of the TTC. -> I used median value of distance ratio.

void computeTTCCamera(std::vector<cv::KeyPoint> &kptsPrev, std::vector<cv::KeyPoint> &kptsCurr, 
                      std::vector<cv::DMatch> kptMatches, double frameRate, double &TTC, cv::Mat *visImg)
{
    // compute distance ratios between all matched keypoints
    vector<double> distRatios; // stores the distance ratios for all keypoints between curr. and prev. frame
    for (auto it1 = kptMatches.begin(); it1 != kptMatches.end() - 1; ++it1)
    { // outer keypoint loop

        // get current keypoint and its matched partner in the prev. frame
        cv::KeyPoint kpOuterCurr = kptsCurr.at(it1->trainIdx);
        cv::KeyPoint kpOuterPrev = kptsPrev.at(it1->queryIdx);

        for (auto it2 = kptMatches.begin() + 1; it2 != kptMatches.end(); ++it2)
        { // inner keypoint loop

            double minDist = 100.0; // min. required distance

            // get next keypoint and its matched partner in the prev. frame
            cv::KeyPoint kpInnerCurr = kptsCurr.at(it2->trainIdx);
            cv::KeyPoint kpInnerPrev = kptsPrev.at(it2->queryIdx);

            // compute distances and distance ratios
            double distCurr = cv::norm(kpOuterCurr.pt - kpInnerCurr.pt);
            double distPrev = cv::norm(kpOuterPrev.pt - kpInnerPrev.pt);

            if (distPrev > std::numeric_limits<double>::epsilon() && distCurr >= minDist)
            { // avoid division by zero

                double distRatio = distCurr / distPrev;
                distRatios.push_back(distRatio);
            }
        } // eof inner loop over all matched kpts
    } //eof outer loop over all matched kpts

    // only continue if list of distance ratios is not empty
    if (distRatios.size() == 0)
    {
        TTC = NAN;
        return;
    }

    // use median distance ratio
    std::sort(distRatios.begin(), distRatios.end()); // sort distRatios
    int medIndex = floor(distRatios.size() / 2.0); //get medindex   ex.)if 3.5 -> 3
    // if even number, get two median values and mean of them. index starts with 0.
    double medDistRatio = distRatios.size() % 2 == 0 ? (distRatios[medIndex - 1] + distRatios[medIndex]) / 2.0 : distRatios[medIndex]; 
    
    double dT = 1 / frameRate;
    TTC = - dT / (1 - medDistRatio);
}

FP.5 Performance Evaluation 1

Look for several examples where you have the impression that the Lidar-based TTC estimate is way off. Once you have found those, describe your observations and provide a sound argumentation why you think this happened.

Since the model of constant-velocity is used, TTC of both Lidar and Camera should decrease. However, as you can see down below, the best performance model from Midterm project (based on matching rate) showed a little increase at frame 35, 89, and 11. This tendency has also been shown for other detectors and descriptors.

Shi-Tomasi detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.6733
Frame 2 12.6142 14.3375
Frame 3 14.091 15.4413
Frame 4 16.6894 12.0075
Frame 5 15.7465 13.2483
Frame 6 12.7835 15.3175
Frame 7 11.9844 14.1036
Frame 8 13.1241 14.3413
Frame 9 13.0241 12.2953
Frame 10 11.1746 14.0712
Frame 11 12.8086 12.0367
Frame 12 8.95978 12.1312
Frame 13 9.96439 12.9306
Frame 14 9.59863 13.4865
Frame 15 8.52157 13.0675
Frame 16 9.51552 12.8883
Frame 17 9.61241 12.9008
Frame 18 8.3988 10.8327

Why sudden Increase? We manually estimated the distance to the rear of the preceding vehicle from a top view perspective of the Lidar points. There is the photo of the cloud of the preceding vehicle's rear part. As you can see in the picture, there are points on the back of the car that are significantly different from other Lidar points. The median value is calculated by x coordinates of these points, which pushed the median value to the back. As a result, the value of medXprev among TTC = medXcurr * dT / (medXprev - medXcurr) was calculated shorter, which led to a larger TTC value.

This can be improved by eliminating the points which is noticeably off, and then calculate median. One of the other ways to solve this problem is to use a bigger shrinkFactor(=0.1~0.2) in order to get more stable and reliable cloud points within interesting bounding box.

FP.6 Performance Evaluation 2

Run the different detector / descriptor combinations and look at the differences in TTC estimation. Find out which methods perform best and also include several examples where camera-based TTC estimation is way off. Describe your observations again and also look into potential reasons.

  1. The use of Harris Detector tended to cause poor performance of TTC using camera.

Harris detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 10.586
Frame 3 14.091 26.2905
Frame 4 16.6894 11.7693
Frame 5 15.7465 -inf
Frame 6 12.7835 12.9945
Frame 7 11.9844 12.2792
Frame 8 13.1241 12.9162
Frame 9 13.0241 nan
Frame 10 11.1746 -inf
Frame 11 12.8086 11.2142
Frame 12 8.95978 12.245
Frame 13 9.96439 13.456
Frame 14 9.59863 5.6061
Frame 15 8.52157 -13.6263
Frame 16 9.51552 6.33866
Frame 17 9.61241 12.5848
Frame 18 8.3988 -inf

Harris detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 10.586
Frame 3 14.091 -10.8522
Frame 4 16.6894 12.1284
Frame 5 15.7465 35.3833
Frame 6 12.7835 13.5907
Frame 7 11.9844 12.2792
Frame 8 13.0241 3.30058
Frame 9 11.1746 20.5862
Frame 10 12.8086 11.8135
Frame 11 8.95978 nan
Frame 12 9.96439 -inf
Frame 13 9.59863 5.6061
Frame 14 8.52157 -13.6263
Frame 15 9.51552 6.6376
Frame 16 9.61241 12.5848
Frame 17 8.3988 -inf

This can be explained by number of keypoints that harris detect. Number of keypoints Harris detected was significantly small, compared to other detectors.

Number of Keypoints

Detector Keypoint #(10 frames - ROI) Keypoint #(Average - ROI)
Harris 17, 14, 18, 21, 26, 43, 18, 31, 26, 34 24.80
Shi-Tomasi 125, 118, 123, 120, 120, 113, 114, 123, 111 ,112 117.90
FAST 149, 152, 150, 155, 149, 149, 156, 150, 138, 143 149.10
BRISK 264, 282, 282, 277, 297, 279, 289, 272, 266, 254 276.20
ORB 92, 102, 106, 113, 109, 125, 130, 129, 127, 128 116.10
AKAZE 166, 157, 161, 155, 163, 164, 173, 175, 177, 179 167.00
SIFT 138, 132, 124, 137, 134, 140, 137, 148, 159, 137 138.60

And this led to small number of keypoints within current bounding box, and this lack of data led to miscalculation of distance ratios between current and previous frame. And as a result, the calculated value of the TTC is not stable.

For example, look at the Euclid distance computation of Harris & BRISK.

[Euclidean distance within frame 2 & 3] 0 0 0 1 1.41421 1.41421

Then, by

double outlierRatio = 0.2;
int idx = floor(euclidDistance.size() * (1 - outlierRatio));

only 4 keypoints remain in the bounding box.

  1. ORB detector, which also has a small number of key points and poor matching rates, also showed poor performance in general.

ORB detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.4733
Frame 2 12.6142 20.5211
Frame 3 14.091 10.5726
Frame 4 15.7465 -inf
Frame 5 11.9844 -inf
Frame 6 13.1241 9.02151
Frame 7 13.0241 22.8248
Frame 8 11.1746 -inf
Frame 9 12.8086 8.76336
Frame 10 8.95978 8.41658
Frame 11 9.96439 8.51436
Frame 12 9.59863 27.8458
Frame 13 8.52157 -inf
Frame 14 9.51552 10.8449
Frame 15 9.61241 9.93894
Frame 16 8.3988 7.62862

Why nan?

if (distPrev > std::numeric_limits<double>::epsilon() && distCurr >= minDist)
{ // avoid division by zero

    double distRatio = distCurr / distPrev;
    distRatios.push_back(distRatio);
}

...
...

// only continue if list of distance ratios is not empty
if (distRatios.size() == 0)
{
    TTC = NAN;
    return;
}
'''
When distance between two keypoints in current frame image is smaller than minDist, size or distRatios becomes zero then returns TTC = nan. 

**Why -inf?**
This occurs when lack of keypoint data, all calculated inliers of euclidean distance between previous and current keypoints are 0. Then median value of Distance ratio becomes 1.

TTC = - dT / (1 - medDistRatio);

Then TTC becomes -inf.

**The best Combinations**

The least "increase of TTC between frame (=error)" was rated as having the best performance. The Shi-Tomasi detector & BRIEF descriptor, which has the smallest value of 2.5984s, was chosen as the best combination.

1. Run Time
Best Performance : FAST + BRIEF

2. Matching Rate
Best Performance : Shi-Tomasi + BRIEF

3. Increase of TTC (=error)
Best Performance : Shi-Tomasi + BRIEF

**TTC of Detector & Descriptor Combinations**

※ Difference between the number of image frame occured due to no observed bounding box using YOLO. This can be fixed by setting greater size of divided cells in pixels.
```    cv::Size size = cv::Size(416, 416);
    // cv::Size size = cv::Size(608, 608);

Harris detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 10.586
Frame 3 14.091 26.2905
Frame 4 16.6894 11.7693
Frame 5 15.7465 -inf
Frame 6 12.7835 12.9945
Frame 7 11.9844 12.2792
Frame 8 13.1241 12.9162
Frame 9 13.0241 nan
Frame 10 11.1746 -inf
Frame 11 12.8086 11.2142
Frame 12 8.95978 12.245
Frame 13 9.96439 13.456
Frame 14 9.59863 5.6061
Frame 15 8.52157 -13.6263
Frame 16 9.51552 6.33866
Frame 17 9.61241 12.5848
Frame 18 8.3988 -inf

Harris detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 10.586
Frame 3 14.091 -10.8522
Frame 4 16.6894 12.1284
Frame 5 15.7465 35.3833
Frame 6 12.7835 13.5907
Frame 7 11.9844 12.2792
Frame 8 13.0241 3.30058
Frame 9 11.1746 20.5862
Frame 10 12.8086 11.8135
Frame 11 8.95978 nan
Frame 12 9.96439 -inf
Frame 13 9.59863 5.6061
Frame 14 8.52157 -13.6263
Frame 15 9.51552 6.6376
Frame 16 9.61241 12.5848
Frame 17 8.3988 -inf

Harris detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 10.586
Frame 3 14.091 -11.4731
Frame 4 16.6894 11.7693
Frame 5 15.7465 35.3833
Frame 6 12.7835 13.5907
Frame 7 11.9844 13.497
Frame 8 13.1241 17.6204
Frame 9 13.0241 nan
Frame 10 11.1746 nan
Frame 11 12.8086 11.2142
Frame 12 8.95978 11.9536
Frame 13 9.96439 13.4327
Frame 14 9.59863 5.85828
Frame 15 8.52157 -12.639
Frame 16 9.51552 6.60338
Frame 17 9.61241 12.5848
Frame 18 8.3988 -inf

Harris detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 9.74953
Frame 2 12.6142 10.586
Frame 3 14.091 nan
Frame 4 16.6894 nan
Frame 5 15.7465 44.9166
Frame 6 12.7835 14.1981
Frame 7 11.9844 12.2
Frame 8 13.0241 nan
Frame 9 11.1746 10.2931
Frame 10 12.8086 23.627
Frame 11 8.95978 11.9536
Frame 12 9.96439 nan
Frame 13 9.59863 nan
Frame 14 8.52157 -25.2781
Frame 15 9.51552 6.87135
Frame 16 9.61241 11.1009
Frame 17 8.3988 -inf

Harris detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.9082
Frame 2 12.6142 11.0081
Frame 3 14.091 -11.4731
Frame 4 15.7465 35.3833
Frame 5 11.9844 13.1905
Frame 6 13.1241 17.6204
Frame 7 13.0241 3.30058
Frame 8 11.1746 20.5862
Frame 9 12.8086 11.7414
Frame 10 8.95978 12.245
Frame 11 9.96439 568.322
Frame 12 9.59863 5.6061
Frame 13 8.52157 -13.6263
Frame 14 9.51552 7.03775
Frame 15 9.61241 12.5848
Frame 16 8.3988 -inf

Shi-Tomasi detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.049
Frame 2 12.6142 12.6968
Frame 3 14.091 14.4793
Frame 4 16.6894 12.7502
Frame 5 15.7465 12.5109
Frame 6 12.7835 15.1536
Frame 7 11.9844 13.1788
Frame 8 13.1241 14.2932
Frame 9 13.0241 12.5494
Frame 10 11.1746 13.3275
Frame 11 12.8086 11.494
Frame 12 8.95978 12.287
Frame 13 9.96439 13.3352
Frame 14 9.59863 12.6532
Frame 15 8.52157 11.8987
Frame 16 9.51552 11.6088
Frame 17 9.61241 12.828
Frame 18 8.3988 11.4955

Shi-Tomasi detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.6733
Frame 2 12.6142 14.3375
Frame 3 14.091 15.4413
Frame 4 16.6894 12.0075
Frame 5 15.7465 13.2483
Frame 6 12.7835 15.3175
Frame 7 11.9844 14.1036
Frame 8 13.1241 14.3413
Frame 9 13.0241 12.2953
Frame 10 11.1746 14.0712
Frame 11 12.8086 12.0367
Frame 12 8.95978 12.1312
Frame 13 9.96439 12.9306
Frame 14 9.59863 13.4865
Frame 15 8.52157 13.0675
Frame 16 9.51552 12.8883
Frame 17 9.61241 12.9008
Frame 18 8.3988 10.8327

Shi-Tomasi detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.1064
Frame 2 12.6142 12.3445
Frame 3 14.091 13.629
Frame 4 16.6894 12.4346
Frame 5 15.7465 13.2534
Frame 6 12.7835 15.917
Frame 7 11.9844 15.0065
Frame 8 13.1241 15.0765
Frame 9 13.0241 11.6077
Frame 10 11.1746 14.0712
Frame 11 12.8086 11.6215
Frame 12 8.95978 12.1102
Frame 13 9.96439 12.6502
Frame 14 9.59863 12.5716
Frame 15 8.52157 10.9042
Frame 16 9.51552 13.1503
Frame 17 9.61241 11.1201
Frame 18 8.3988 11.1899

Shi-Tomasi detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.0182
Frame 2 12.6142 12.5187
Frame 3 14.091 14.2771
Frame 4 15.7465 12.7336
Frame 5 12.7835 13.4761
Frame 6 11.9844 13.4548
Frame 7 13.1241 12.0155
Frame 8 13.0241 12.5073
Frame 9 11.1746 14.0134
Frame 10 12.8086 11.7837
Frame 11 8.95978 12.9405
Frame 12 9.96439 13.6254
Frame 13 9.59863 12.241
Frame 14 8.52157 12.6641
Frame 15 9.51552 10.3773
Frame 16 9.61241 13.1247
Frame 17 8.3988 10.6119
Frame 18 6.915 1.00815

Shi-Tomasi detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.0746
Frame 2 12.6142 14.6618
Frame 3 14.091 17.181
Frame 4 16.6894 13.0192
Frame 5 15.7465 13.3479
Frame 6 12.7835 14.1323
Frame 7 11.9844 13.935
Frame 8 13.1241 14.5085
Frame 9 13.0241 12.2319
Frame 10 11.1746 13.7396
Frame 11 12.8086 12.3562
Frame 12 8.95978 12.269
Frame 13 9.96439 12.8498
Frame 14 9.59863 12.7102
Frame 15 8.52157 12.3292
Frame 16 9.51552 13.1068
Frame 17 9.61241 11.5312
Frame 18 8.3988 10.9222

FAST detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 12.313
Frame 2 12.6142 13.5384
Frame 3 14.091 13.9572
Frame 4 16.6894 13.0212
Frame 5 15.7465 99.5628
Frame 6 11.9844 13.1858
Frame 7 13.1241 11.9077
Frame 8 13.0241 13.4042
Frame 9 11.1746 14.3255
Frame 10 12.8086 12.3427
Frame 11 8.95978 13.5201
Frame 12 9.96439 12.7784
Frame 13 9.59863 12.879
Frame 14 8.52157 13.0655
Frame 15 9.51552 12.9162
Frame 16 9.61241 11.0958
Frame 17 8.3988 12.3793

FAST detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 12.5
Frame 2 12.6142 13.0256
Frame 3 14.091 13.9811
Frame 4 15.7465 331.161
Frame 5 12.7835 13.6183
Frame 6 11.9844 12.13
Frame 7 13.1241 12.0007
Frame 8 13.0241 14.808
Frame 9 11.1746 13.1088
Frame 10 12.8086 14.7288
Frame 11 8.95978 11.7494
Frame 12 9.96439 12.8743
Frame 13 9.59863 12.5199
Frame 14 8.52157 12.1879
Frame 15 9.51552 12.5159
Frame 16 9.61241 9.82934
Frame 17 8.3988 12.5588

FAST detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 10.7912
Frame 2 12.6142 12.7571
Frame 3 14.091 20.9306
Frame 4 15.7465 244.85
Frame 5 11.9844 13.3198
Frame 6 13.0241 13.4602
Frame 7 11.1746 13.5653
Frame 8 12.8086 14.558
Frame 9 8.95978 13.6506
Frame 10 9.96439 12.8506
Frame 11 9.59863 12.1368
Frame 12 8.52157 13.9397
Frame 13 9.51552 12.2753
Frame 14 9.61241 11.7698
Frame 15 8.3988 13.2302

FAST detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 8.44411
Frame 2 12.6142 27.8578
Frame 3 14.091 12.1376
Frame 4 15.7465 13.0459
Frame 5 12.7835 12.6664
Frame 6 11.9844 12.1002
Frame 7 13.1241 11.7849
Frame 8 13.0241 13.1008
Frame 9 11.1746 13.8308
Frame 10 12.8086 12.5157
Frame 11 8.95978 13.3321
Frame 12 9.96439 12.8514
Frame 13 9.59863 12.1368
Frame 14 8.52157 12.0969
Frame 15 9.51552 12.2808
Frame 16 9.61241 11.8476
Frame 17 8.3988 12.6702

FAST detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 12.3317
Frame 2 12.6142 13.4194
Frame 3 14.091 22.7505
Frame 4 15.7465 199.157
Frame 5 12.7835 13.9137
Frame 6 11.9844 12.2781
Frame 7 13.1241 12.6536
Frame 8 13.0241 13.3673
Frame 9 11.1746 14.0112
Frame 10 12.8086 15.8801
Frame 11 8.95978 12.7731
Frame 12 9.96439 12.9215
Frame 13 9.59863 12.8223
Frame 14 8.52157 12.5997
Frame 15 9.51552 12.9069
Frame 16 9.61241 10.5264
Frame 17 8.3988 12.0819

BRISK detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.4528
Frame 2 12.6142 36.8812
Frame 3 14.091 21.4969
Frame 4 16.6894 18.4003
Frame 5 15.7465 30.382
Frame 6 11.9844 21.0793
Frame 7 13.1241 23.1712
Frame 8 13.0241 20.4435
Frame 9 11.1746 16.2296
Frame 10 12.8086 15.1292
Frame 11 8.95978 15.3676
Frame 12 9.96439 16.8161
Frame 13 9.59863 15.91
Frame 14 8.52157 16.2868
Frame 15 9.51552 14.7862
Frame 16 9.61241 12.2143
Frame 17 8.3988 16.2886

BRISK detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.8273
Frame 2 12.6142 19.013
Frame 3 14.091 16.8644
Frame 4 16.6894 20.3782
Frame 5 15.7465 15.4176
Frame 6 11.9844 17.0509
Frame 7 13.1241 18.824
Frame 8 13.0241 21.0641
Frame 9 11.1746 17.3253
Frame 10 12.8086 15.8866
Frame 11 8.95978 16.0764
Frame 12 9.96439 17.4274
Frame 13 9.59863 15.0951
Frame 14 8.52157 16.0729
Frame 15 9.51552 15.9585
Frame 16 9.61241 10.8117
Frame 17 8.3988 15.6634

BRISK detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.1922
Frame 2 12.6142 30.0422
Frame 3 14.091 17.943
Frame 4 16.6894 16.6317
Frame 5 15.7465 22.2484
Frame 6 12.7835 34.0073
Frame 7 11.9844 16.1467
Frame 8 13.1241 17.5377
Frame 9 13.0241 21.8078
Frame 10 11.1746 12.8042
Frame 11 12.8086 15.3598
Frame 12 8.95978 14.7782
Frame 13 9.96439 13.4943
Frame 14 9.59863 20.4401
Frame 15 8.52157 13.3384
Frame 16 9.51552 14.2821
Frame 17 9.61241 9.70105
Frame 18 8.3988 15.1448

BRISK detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 22.3338
Frame 2 12.6142 24.1462
Frame 3 14.091 17.3415
Frame 4 15.7465 45.6935
Frame 5 12.7835 19.6499
Frame 6 11.9844 20.719
Frame 7 13.1241 19.7292
Frame 8 13.0241 19.3079
Frame 9 11.1746 13.1305
Frame 10 12.8086 16.1616
Frame 11 8.95978 14.6918
Frame 12 9.96439 15.3799
Frame 13 9.59863 15.9249
Frame 14 8.52157 14.4594
Frame 15 9.51552 13.3721
Frame 16 9.61241 12.1469
Frame 17 8.3988 13.8154

BRISK detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.7651
Frame 2 12.6142 17.4898
Frame 3 14.091 16.9848
Frame 4 15.7465 24.0727
Frame 5 12.7835 18.8188
Frame 6 11.9844 16.4298
Frame 7 13.1241 16.8197
Frame 8 13.0241 16.3258
Frame 9 11.1746 16.2161
Frame 10 12.8086 12.0368
Frame 11 8.95978 11.8363
Frame 12 9.96439 13.6835
Frame 13 9.59863 11.48
Frame 14 8.52157 13.3989
Frame 15 9.51552 11.0579
Frame 16 9.61241 11.4685
Frame 17 8.3988 13.1412

ORB detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 19.6003
Frame 2 12.6142 19.6382
Frame 3 14.091 17.7441
Frame 4 15.7465 156.224
Frame 5 11.9844 12.9706
Frame 6 13.1241 10.911
Frame 7 13.0241 -inf
Frame 8 11.1746 -inf
Frame 9 12.8086 7.52121
Frame 10 8.95978 -inf
Frame 11 9.96439 22.4629
Frame 12 9.59863 35.3185
Frame 13 8.52157 13.594
Frame 14 9.51552 14.2951
Frame 15 9.61241 23.6651
Frame 16 8.3988 46.6458

ORB detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 16.5073
Frame 2 12.6142 -inf
Frame 3 14.091 89.2313
Frame 4 15.7465 24.0009
Frame 5 11.9844 -inf
Frame 6 13.1241 -inf
Frame 7 13.0241 -inf
Frame 8 11.1746 12.8303
Frame 9 12.8086 24.9891
Frame 10 8.95978 16.9677
Frame 11 9.96439 -inf
Frame 12 9.59863 21.8391
Frame 13 8.52157 164.864
Frame 14 9.51552 13.0005
Frame 15 9.61241 16.8713
Frame 16 8.3988 23.002

ORB detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 19.4709
Frame 2 12.6142 -inf
Frame 3 14.091 23.9536
Frame 4 15.7465 29.6544
Frame 5 11.9844 -inf
Frame 6 13.1241 -inf
Frame 7 13.0241 -inf
Frame 8 11.1746 -inf
Frame 9 12.8086 9.35938
Frame 10 8.95978 -inf
Frame 11 9.96439 -inf
Frame 12 9.59863 177.585
Frame 13 8.52157 -inf
Frame 14 9.51552 26.3083
Frame 15 9.61241 16.6667
Frame 16 8.3988 -inf

ORB detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.4733
Frame 2 12.6142 20.5211
Frame 3 14.091 10.5726
Frame 4 15.7465 -inf
Frame 5 11.9844 -inf
Frame 6 13.1241 9.02151
Frame 7 13.0241 22.8248
Frame 8 11.1746 -inf
Frame 9 12.8086 8.76336
Frame 10 8.95978 8.41658
Frame 11 9.96439 8.51436
Frame 12 9.59863 27.8458
Frame 13 8.52157 -inf
Frame 14 9.51552 10.8449
Frame 15 9.61241 9.93894
Frame 16 8.3988 7.62862

ORB detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 16.726
Frame 2 12.6142 19.2415
Frame 3 14.091 12.0957
Frame 4 15.7465 502.13
Frame 5 11.9844 -inf
Frame 6 13.1241 11.3168
Frame 7 13.0241 -inf
Frame 8 11.1746 -inf
Frame 9 12.8086 7.44098
Frame 10 8.95978 -inf
Frame 11 9.96439 9.91242
Frame 12 9.59863 35.6903
Frame 13 8.52157 33.3507
Frame 14 9.51552 12.1158
Frame 15 9.61241 21.392
Frame 16 8.3988 20.1371

AKAZE detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 16.8563
Frame 2 12.6142 18.4929
Frame 3 14.091 16.3504
Frame 4 16.6894 17.5351
Frame 5 15.7465 18.6746
Frame 6 12.7835 20.3714
Frame 7 11.9844 21.7319
Frame 8 13.1241 17.9099
Frame 9 13.0241 22.2591
Frame 10 11.1746 14.6515
Frame 11 12.8086 15.0257
Frame 12 8.95978 15.3756
Frame 13 9.96439 13.1185
Frame 14 9.59863 12.7676
Frame 15 8.52157 13.6149
Frame 16 9.51552 13.0152
Frame 17 9.61241 11.4466
Frame 18 8.3988 10.6396

AKAZE detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 15.4228
Frame 2 12.6142 19.1145
Frame 3 14.091 15.181
Frame 4 15.7465 18.8963
Frame 5 11.9844 21.1049
Frame 6 13.1241 19.3563
Frame 7 13.0241 20.7751
Frame 8 11.1746 14.1156
Frame 9 12.8086 14.507
Frame 10 8.95978 12.4654
Frame 11 9.96439 13.3461
Frame 12 9.59863 13.3441
Frame 13 8.52157 11.9791
Frame 14 9.51552 11.947
Frame 15 9.61241 10.8779
Frame 16 8.3988 10.6942

AKAZE detector & ORB descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.0321
Frame 2 12.6142 16.64
Frame 3 14.091 14.7873
Frame 4 16.6894 15.3607
Frame 5 15.7465 18.2013
Frame 6 12.7835 16.9835
Frame 7 11.9844 18.621
Frame 8 13.1241 17.178
Frame 9 13.0241 18.9619
Frame 10 11.1746 13.7636
Frame 11 12.8086 13.7961
Frame 12 8.95978 12.7123
Frame 13 9.96439 13.0723
Frame 14 9.59863 14.2072
Frame 15 8.52157 14.4576
Frame 16 9.51552 12.5881
Frame 17 9.61241 11.6888
Frame 18 8.3988 9.8149

AKAZE detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 12.7377
Frame 2 12.6142 17.5354
Frame 3 14.091 16.1137
Frame 4 16.6894 16.713
Frame 5 15.7465 18.5138
Frame 6 12.7835 21.009
Frame 7 11.9844 20.135
Frame 8 13.1241 16.9118
Frame 9 13.0241 21.0994
Frame 10 11.1746 14.9509
Frame 11 12.8086 14.9483
Frame 12 8.95978 16.5268
Frame 13 9.96439 14.023
Frame 14 9.59863 12.5126
Frame 15 8.52157 12.6712
Frame 16 9.51552 11.866
Frame 17 9.61241 11.3796
Frame 18 8.3988 10.6609

AKAZE detector & AKAZE descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.7813
Frame 2 12.6142 18.2759
Frame 3 14.091 15.2954
Frame 4 15.7465 18.578
Frame 5 12.7835 18.5209
Frame 6 11.9844 17.6798
Frame 7 13.1241 18.4169
Frame 8 13.0241 18.8738
Frame 9 11.1746 14.3961
Frame 10 12.8086 14.7002
Frame 11 8.95978 14.0471
Frame 12 9.96439 13.1209
Frame 13 9.59863 14.0193
Frame 14 8.52157 12.7913
Frame 15 9.51552 12.3797
Frame 16 9.61241 10.9403
Frame 17 8.3988 10.4431

AKAZE detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.275
Frame 2 12.6142 17.924
Frame 3 14.091 16.1258
Frame 4 15.7465 19.5558
Frame 5 12.7835 19.9529
Frame 6 11.9844 17.8047
Frame 7 13.1241 17.777
Frame 8 13.0241 18.4054
Frame 9 11.1746 13.9839
Frame 10 12.8086 14.1883
Frame 11 8.95978 14.3416
Frame 12 9.96439 12.9535
Frame 13 9.59863 13.3441
Frame 14 8.52157 13.2259
Frame 15 9.51552 11.9771
Frame 16 9.61241 11.04
Frame 17 8.3988 10.2489

SIFT detector & BRISK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 16.3605
Frame 2 12.6142 18.1968
Frame 3 14.091 21.0449
Frame 4 15.7465 21.3614
Frame 5 12.7835 16.0845
Frame 6 11.9844 20.3296
Frame 7 13.0241 17.3057
Frame 8 11.1746 15.38
Frame 9 12.8086 14.1564
Frame 10 8.95978 12.5129
Frame 11 9.96439 13.0106
Frame 12 9.59863 12.0078
Frame 13 8.52157 10.1541
Frame 14 9.51552 10.6313
Frame 15 9.61241 10.9394
Frame 16 8.3988 13.1621

SIFT detector & BRIEF descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 13.2917
Frame 2 12.6142 14.8783
Frame 3 14.091 16.2329
Frame 4 15.7465 20.9343
Frame 5 12.7835 16.6981
Frame 6 11.9844 18.3806
Frame 7 13.1241 17.5379
Frame 8 13.0241 14.234
Frame 9 11.1746 12.5144
Frame 10 12.8086 14.4165
Frame 11 8.95978 12.6976
Frame 12 9.96439 13.1667
Frame 13 9.59863 13.0787
Frame 14 8.52157 11.899
Frame 15 9.51552 9.74405
Frame 16 9.61241 11.2425
Frame 17 8.3988 12.7964

SIFT detector & FREAK descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 16.0812
Frame 2 12.6142 17.7462
Frame 3 14.091 17.4939
Frame 4 15.7465 21.2313
Frame 5 12.7835 15.6675
Frame 6 11.9844 21.2261
Frame 7 13.1241 17.3051
Frame 8 13.0241 17.3556
Frame 9 11.1746 13.8167
Frame 10 12.8086 15.1919
Frame 11 8.95978 13.2843
Frame 12 9.96439 11.8665
Frame 13 9.59863 15.199
Frame 14 8.52157 11.7308
Frame 15 9.51552 9.73895
Frame 16 9.61241 10.0683
Frame 17 8.3988 14.2603

SIFT detector & SIFT descriptor

Image Frame TTC Lidar(s) TTC Camera(s)
Frame 1 12.5156 14.738
Frame 2 12.6142 15.8621
Frame 3 14.091 15.4252
Frame 4 15.7465 18.44
Frame 5 12.7835 15.273
Frame 6 11.9844 16.1902
Frame 7 13.1241 18.5833
Frame 8 13.0241 14.4655
Frame 9 11.1746 13.0114
Frame 10 12.8086 13.5342
Frame 11 8.95978 14.5151
Frame 12 9.96439 12.7747
Frame 13 9.59863 12.9075
Frame 14 8.52157 10.5037
Frame 15 9.51552 10.5037
Frame 16 9.61241 9.52376
Frame 17 8.3988 11.4407

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

1 star

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages