The goals / steps of this project are the following:
- Perform a Histogram of Oriented Gradients (HOG) feature extraction on a labeled training set of images and train a classifier Linear SVM classifier
- Optionally, you can also apply a color transform and append binned color features, as well as histograms of color, to your HOG feature vector.
- Note: for those first two steps don't forget to normalize your features and randomize a selection for training and testing.
- Implement a sliding-window technique and use your trained classifier to search for vehicles in images.
- Run your pipeline on a video stream (start with the test_video.mp4 and later implement on full project_video.mp4) and create a heat map of recurring detections frame by frame to reject outliers and follow detected vehicles.
- Estimate a bounding box for vehicles detected.
Rubric Points
Here I will consider the rubric points individually and describe how I addressed each point in my implementation.
1. Provide a Writeup / README that includes all the rubric points and how you addressed each one. You can submit your writeup as markdown or pdf. Here is a template writeup for this project you can use as a guide and a starting point.
You're reading it!
1. Explain how (and identify where in your code) you extracted HOG features from the training images.
The code for this step is contained in the first code cell of the IPython notebook (or in lines # through # of the file called some_file.py).
I started by reading in all the vehicle and non-vehicle images. Here is an example of one of each of the vehicle and non-vehicle classes:
I then explored different color spaces and different skimage.hog() parameters (orientations, pixels_per_cell, and cells_per_block). I grabbed random images from each of the two classes and displayed them to get a feel for what the skimage.hog() output looks like.
Here is an example using the YCrCb color space and HOG parameters of orientations=9, pixels_per_cell=(8, 8) and cells_per_block=(2, 2):
I tried various combinations of parameters and found that this combination is reasonably fast and produces good results. Apart from HOG, spatial binning and color histograms on all three channels were used as shown in Cell 2 of the Jupyter notebook.
3. Describe how (and identify where in your code) you trained a classifier using your selected HOG features (and color features if you used them).
I trained a linear SVM using two separate classes of vehicle images and non-vehicle images. There were more than 8500 images in each class totalling more than 17500 training images.
The images were loaded, then feature extraction as described above, was done on each image (in Cell 2).
The extracted features were shuffled and split in 80:20 ratio to form two datasets train (80%) and test (20%). The features were then scaled to zero mean and unit variance using the scikit StandardScaler() in Cell 5. The SVC (Support Vector Classifier) was trained in Cell 5 and resulted in a very high accuracy of ~99%.
1. Describe how (and identify where in your code) you implemented a sliding window search. How did you decide what scales to search and how much to overlap windows?
I decided to make use of a single function that's able to both extract features and make predictions. Search area for vehicles was focused in the lower part of the image using an overlapping sliding window search and to make the search faster. The window size is 64 pixels, with 8 cells and 8 pix per cell. At each slide the windows move by 2 cells either to the right or to the bottom. For efficiency,the feature extraction is done only once for the image and then sliding window operates only on part of the image. The detection is made more robust by using different scales for accommodating cars at long and short distances.
2. Show some examples of test images to demonstrate how your pipeline is working. What did you do to optimize the performance of your classifier?
Ultimately I searched on four scales using YCrCb 3-channel HOG features plus spatially binned color and histograms of color in the feature vector, which provided a nice result. Here are some example images:
1. Provide a link to your final video output. Your pipeline should perform reasonably well on the entire project video (somewhat wobbly or unstable bounding boxes are ok as long as you are identifying the vehicles most of the time with minimal false positives.)
Here's a link to my video result
2. Describe how (and identify where in your code) you implemented some kind of filter for false positives and some method for combining overlapping bounding boxes.
I recorded the positions of positive detections in each frame of the video. From the positive detections I created a heatmap and then thresholded that map to identify vehicle positions. I then used scipy.ndimage.measurements.label() to identify individual blobs in the heatmap. I then assumed each blob corresponded to a vehicle. I constructed bounding boxes to cover the area of each blob detected.
Here's an example result showing the heatmap from a series of frames of video, the result of scipy.ndimage.measurements.label() and the bounding boxes then overlaid on the last frame of video:
Here is the output of scipy.ndimage.measurements.label() on the integrated heatmap from all six frames:
1. Briefly discuss any problems / issues you faced in your implementation of this project. Where will your pipeline likely fail? What could you do to make it more robust?
Pipeline is seen performing badly in big shadowy areas despite using YCrCb channel. Smarter thresholding can be utilized where an objects that persist in consecutive frames (unlike shadows and some false positives) will be tracked smoothly. Averaging results over past 'n' frames would help to make a smoother video. Use of multi-scaled sliding windows also helped in making pipeline more robust in identifying distant and nearby cars.