About Object Heading constraint

[ihaohe]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

[JinglinZhan]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object $O$ in a scene $S$, we first find out all the ground-truth objects with the same category as $O$ in $S$ and generate the heading angle collection ${\theta_c}$. If ${\theta_c}$ is not empty, we use Eq.4 to choose most suitable $\theta$ for $O$: Calculate the total cosine similarity between $\theta$ and all elements in ${\theta_c}$ and choose the one with highest similarity value.

[ihaohe]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object O in a scene S , we first find out all the ground-truth objects with the same category as O in S and generate the heading angle collection θ c . If θ c is not empty, we use Eq.4 to choose most suitable θ for O : Calculate the total cosine similarity between θ and all elements in θ c and choose the one with highest similarity value.

Thanks for your reply. As you said “Calculate the total cosine similarity between $\theta$ and all elements in $\theta_c$ ", I know what the $\theta_c$ means but confused with the symbol $\theta$, Is this $\theta$ the same as Eq.3 or it comes from sampling in circle. Can you explain in detail？

[JinglinZhan]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object O in a scene S , we first find out all the ground-truth objects with the same category as O in S and generate the heading angle collection θ c . If θ c is not empty, we use Eq.4 to choose most suitable θ for O : Calculate the total cosine similarity between θ and all elements in θ c and choose the one with highest similarity value.

Thanks for your reply. As you said “Calculate the total cosine similarity between θ and all elements in θ c ", I know what the θ c means but confused with the symbol θ , Is this θ the same as Eq.3 or it comes from sampling in circle. Can you explain in detail？

Eq. 2 and 3 constrain the distance and observing angle from the ego vehicle of the sampled objects. In other words, the sampled objects can only rotate along the z-axis. The variable $\theta$ represents the heading angle of these rotated objects.

[ihaohe]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object O in a scene S , we first find out all the ground-truth objects with the same category as O in S and generate the heading angle collection θ c . If θ c is not empty, we use Eq.4 to choose most suitable θ for O : Calculate the total cosine similarity between θ and all elements in θ c and choose the one with highest similarity value.

Thanks for your reply. As you said “Calculate the total cosine similarity between θ and all elements in θ c ", I know what the θ c means but confused with the symbol θ , Is this θ the same as Eq.3 or it comes from sampling in circle. Can you explain in detail？

Eq. 2 and 3 constrain the distance and observing angle from the ego vehicle of the sampled objects. In other words, the sampled objects can only rotate along the z-axis. The variable θ represents the heading angle of these rotated objects.

Thanks for your explain. Another question is here, In Sec3.2.2, you said "If there is no object in the scene with the same category, we simply choose the most frequent heading for our object.", dose the most frequent heading mean the heading of the same category in total dataset or all the object heading in current scene?

[JinglinZhan]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object O in a scene S , we first find out all the ground-truth objects with the same category as O in S and generate the heading angle collection θ c . If θ c is not empty, we use Eq.4 to choose most suitable θ for O : Calculate the total cosine similarity between θ and all elements in θ c and choose the one with highest similarity value.

Thanks for your reply. As you said “Calculate the total cosine similarity between θ and all elements in θ c ", I know what the θ c means but confused with the symbol θ , Is this θ the same as Eq.3 or it comes from sampling in circle. Can you explain in detail？

Eq. 2 and 3 constrain the distance and observing angle from the ego vehicle of the sampled objects. In other words, the sampled objects can only rotate along the z-axis. The variable θ represents the heading angle of these rotated objects.

Thanks for your explain. Another question is here, In Sec3.2.2, you said "If there is no object in the scene with the same category, we simply choose the most frequent heading for our object.", dose the most frequent heading mean the heading of the same category in total dataset or all the object heading in current scene?

It means the heading of the same category in total dataset

[ihaohe]

Thanks for your interesting work. I've some questions about The Eq 4. What's the theta meaning of Eq 4 and how to generate the theta. In Fig2 Object position stage, there're many possible locations to select, how dose these locations be generated ?

When inserting an object O in a scene S , we first find out all the ground-truth objects with the same category as O in S and generate the heading angle collection θ c . If θ c is not empty, we use Eq.4 to choose most suitable θ for O : Calculate the total cosine similarity between θ and all elements in θ c and choose the one with highest similarity value.

Thanks for your reply. As you said “Calculate the total cosine similarity between θ and all elements in θ c ", I know what the θ c means but confused with the symbol θ , Is this θ the same as Eq.3 or it comes from sampling in circle. Can you explain in detail？

Eq. 2 and 3 constrain the distance and observing angle from the ego vehicle of the sampled objects. In other words, the sampled objects can only rotate along the z-axis. The variable θ represents the heading angle of these rotated objects.

Thanks for your explain. Another question is here, In Sec3.2.2, you said "If there is no object in the scene with the same category, we simply choose the most frequent heading for our object.", dose the most frequent heading mean the heading of the same category in total dataset or all the object heading in current scene?

It means the heading of the same category in total dataset

Thanks. Wishing you a successful thesis soon!