As I read the collision rate computating code in evaluation.py , I find the following codes in line 127-129, which seem to mean that you only evaluate the first generated trajectory in all 20 output trajectories by selecting [0,0] in the prediction_dict[t][node].
I cannot understand this , to my understanding, all the 20 guesses or the best guesses should be evaluated, instead of selecting the index 0 outputs.
Thank you for your great work!
As I read the collision rate computating code in evaluation.py , I find the following codes in line 127-129, which seem to mean that you only evaluate the first generated trajectory in all 20 output trajectories by selecting [0,0] in the prediction_dict[t][node]. I cannot understand this , to my understanding, all the 20 guesses or the best guesses should be evaluated, instead of selecting the index 0 outputs.
col_joint = compute_col(prediction_dict[t][node][0,0], prediction_joint[idx_neighbors], num_interp=num_interp).astype(float) col_cross = compute_col(prediction_dict[t][node][0,0], futures_joint[idx_neighbors], num_interp=num_interp).astype(float) col_truth = compute_col(futures_dict[t][node], futures_joint[idx_neighbors], num_interp=num_interp)